EFFECTS OF HOMOGENEOUS CHARGE COMPRESSION IGNITION (HCCI)
CONTROL STRATEGIES ON PARTICULATE EMISSIONS OF ETHANOL FUEL
A DISSERTATION
SUBMITTED TO THE FACULTY OF THE GRADUATE SCHOOL
OF THE UNIVERSITY OF MINNESOTA
BY
LUKE FRANKLIN
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
Professor David B. Kittelson, Adviser
December 2010
i
© Luke Franklin 2010
ii
Acknowledgements
This thesis represents a five year continuation of the college years so fondly
remembered by most students. I would like to acknowledge the many people with
which I spent that time because of their contributions as friends, family members, and
colleagues. I do not see the technical accomplishments presented here as something I
will look back upon when reminiscing. But I will undoubtedly remember each of the
kind and intelligent people I was lucky enough to encounter throughout the process.
i
Abstract
This thesis presents a systematic investigation into the formation of particulate matter in
homogeneous charge compression ignition (HCCI) engines. These engines are
representative of the emerging generation of low sooting engine technology. Early
research in the field concluded that engines operating with this combustion strategy
could offer Diesel like efficiency while simultaneously reducing emissions of
particulate matter and the oxides of nitrogen to nearly negligible levels. While
quantification of gas phase emissions has changed little through modern regulatory
history, the metrics defining particulate emissions and the state of understanding of the
research community are rapidly evolving. Advances in technology for characterizing
particulate emissions from spark ignition and compression ignition engines have been
applied to HCCI emissions and the results indicate the production of significant
quantities, by both number and mass, of particles from the HCCI combustion strategy.
A relationship has been identified between in-cylinder behavior, and both gaseous and
particulate emissions. It has been shown to be valid for 2 different fuels and multiple
engine loads. Characteristics of the particulate matter suggest it is formed via gas to
particle conversion, or nucleation, of the lighter distillates from the engines lubricating
oil.
ii
Table of Contents
List of Tables .................................................................................................................... v
List of Figures.................................................................................................................. vi
Nomenclature ................................................................................................................... x
Chapter 1 Introduction .................................................................................................. 1
1.1
Motivation ........................................................................................................ 1
1.2
Statement of Problem ....................................................................................... 2
1.3
Significance ...................................................................................................... 2
1.4
Organization ..................................................................................................... 3
Chapter 2 Background................................................................................................... 5
2.1
Engine Fundamentals ....................................................................................... 5
2.2
Homogeneous Charge Compression Ignition................................................... 9
2.2.1
Historical Perspective ............................................................................. 14
2.2.2
Current Relevant Literature .................................................................... 16
2.3
Emissions........................................................................................................ 24
2.3.1
Spark Ignition Emissions........................................................................ 25
2.3.2
Compression Ignition Emissions............................................................ 27
2.3.3
Homogeneous Charge Compression Ignition Emissions ....................... 28
Chapter 3 PM Emissions Instrumentation................................................................... 37
3.1
Size Distribution Characterization ................................................................. 37
3.1.1
CPC......................................................................................................... 37
3.1.2
SMPS...................................................................................................... 41
3.1.3
EEPS....................................................................................................... 44
3.1.4
TDMA .................................................................................................... 45
3.2
Dilution........................................................................................................... 46
Chapter 4 Preliminary Modeling................................................................................. 52
Chapter 5 Experimental Apparatus ............................................................................. 64
5.1
Multi-cylinder Test Engine............................................................................. 64
5.1.1
Intake Manifold ...................................................................................... 65
5.1.1.1 Fuel Injection...................................................................................... 66
5.1.1.2 EGR .................................................................................................... 67
5.1.1.3 Thermal Management......................................................................... 68
Chapter 6 Effects of Intake Temperature on Emissions From an Ethanol Fueled HCCI
Engine
73
6.1
Experimental Procedure ................................................................................. 74
6.2
Results and Discussion ................................................................................... 76
6.2.1
Combustion Analysis.............................................................................. 77
6.2.2
Emissions Analysis................................................................................. 83
6.3
Conclusions .................................................................................................... 91
Chapter 7 The Effect of EGR on Emissions in an Ethanol Fueled HCCI Engine ...... 93
7.1
Experimental Procedure ................................................................................. 93
7.2
Results and Discussion ................................................................................... 95
7.2.1
Combustion Analysis.............................................................................. 95
7.2.2
Emissions Analysis............................................................................... 100
iii
7.3
Conclusions .................................................................................................. 108
Chapter 8 The Effects Fuel Blending on Emissions in an Ethanol and Hydrogen
Fueled HCCI Engine. ................................................................................................... 110
8.1
Experimental................................................................................................. 111
8.2
Results and Discussion ................................................................................. 113
8.2.1
Combustion Analysis............................................................................ 113
8.2.2
Emissions Analysis............................................................................... 118
8.3
Pure Hydrogen HCCI ................................................................................... 128
8.3.1
Experimental......................................................................................... 129
8.3.2
Combustion Analysis............................................................................ 130
8.3.3
Emissions Analysis............................................................................... 133
8.4
Conclusions .................................................................................................. 139
Chapter 9 Advanced Characterization Techniques for Emissions from an Ethanol
Fueled HCCI Engine .................................................................................................... 141
9.1
TDMA Experiments ..................................................................................... 141
9.1.1
Experimental......................................................................................... 142
9.1.2
Results and Discussion ......................................................................... 144
9.1.3
Conclusions .......................................................................................... 155
9.2
FTIR Data..................................................................................................... 156
9.2.1
Experimental......................................................................................... 157
9.2.2
Results and Discussion ......................................................................... 158
9.2.3
Conclusions .......................................................................................... 163
Chapter 10 Summary and Conclusions ....................................................................... 165
10.1 HCCI Combustion ........................................................................................ 165
10.2 HCCI Emissions ........................................................................................... 166
Bibliography ................................................................................................................. 172
iv
List of Tables
Table 1: Physical characteristics of test engine .............................................................. 52
Table 2: Combustion properties of ethanol and hydrogen fuels..................................... 53
Table 3: λ, IMEP, and rated power relationship for thermal test conditions ................. 56
Table 4: λ, IMEP, and rated power relationship for hydrogen test conditions .............. 59
Table 5: Summary of EtOH HCCI peak pressures with EGR, *indicates misfire ......... 61
Table 6: Ethanol Fuel Composition................................................................................ 67
Table 7: Thermal Management Test Conditions ............................................................ 75
Table 8: Summary of combustion properties, ethanol HCCI with varying intake
temperature, 1500 RPM, 3 loads .................................................................................... 81
Table 9: Test conditions for ethanol HCCI with varying EGR experiments ................. 94
Table 10: Summary of combustion properties, ethanol HCCI with varying EGR rate,
1500 RPM, 3 loads ......................................................................................................... 98
Table 11: Fuel Blending Test Conditions..................................................................... 112
Table 12: Summary of combustion properties, ethanol HCCI with supplemental
hydrogen fueling, 1500 RPM, 3 loads.......................................................................... 116
Table 13: Hydrogen fueled HCCI test conditions ........................................................ 130
Table 14: Summary of combustion properties, hydrogen HCCI with varying intake
temperature, 1500 RPM, 54 Nm Load ......................................................................... 132
Table 15: Engine operating parameters tested in TDMA analysis of ethanol HCCI
combustion ................................................................................................................... 143
Table 16: TDMA bias error data .................................................................................. 144
Table 17: Conventional gas analyzer descriptions ....................................................... 158
Table 18: Chemical Species Examined via FTIR Spectroscopy .................................. 159
Table 19: Average ethanol HCCI exhaust gas composition as measured by conventional
gas analyzers................................................................................................................. 161
Table 20: Ratio of ethanol HCCI exhaust gas measurements made via FTIR compared
with those collected through conventional gas analysis............................................... 162
Table 21: Correlation matrix relating emissions to combustion properties in fully
premixed HCCI combustion of ethanol and hydrogen................................................. 170
v
List of Figures
Figure 1: p-v diagram of the ideal Otto cycle................................................................... 6
Figure 2: p-v diagram of ideal the Diesel cycle................................................................ 6
Figure 3: Charge path in a 4 stroke HCCI engine .......................................................... 10
Figure 4: p-v diagram of the ideal HCCI cycle .............................................................. 11
Figure 5: Cetane number and octane number relationship (Stone, 1999) ...................... 13
Figure 6: Specific heat (cP) of primary exhaust gas components and air ....................... 19
Figure 7: Regions of soot and NOX formation in combustion systems.......................... 30
Figure 8: Butanol CPC particle growth section.............................................................. 39
Figure 9: Water CPC particle growth section................................................................. 40
Figure 10: DMA flow schematic.................................................................................... 42
Figure 11: TDMA Apparatus ......................................................................................... 46
Figure 12: 2 Stage Micro-Dilution System ................................................................... 47
Figure 13: PM variation with stage one dilution air temperature................................... 49
Figure 14: Mean exit temperature profiles along the length of the dilution tunnel, varied
wall temperature, TIn = 50°C, Air flowrate = 80 slpm ................................................... 50
Figure 15: Sensitivity of PM formation to dilution tunnel wall temperature ................. 51
Figure 16: Cylinder pressure traces of simulated HCCI combustion of hydrogen fuel
with λ=2, intake temp. of 355 K, and engine speed of 1000 rpm .................................. 54
Figure 17: Cylinder pressure traces of simulated HCCI combustion of ethanol fuel with
λ=3, intake temp. of 400 K, and engine speed of 1000 rpm .......................................... 55
Figure 18: Pressure vs. CAD at 5 intake temperatures for each of 4 lambda ranges,
EtOH fuel, 1000 RPM .................................................................................................... 57
Figure 19: Pressure vs. CAD with varying hydrogen proportions for each of 4 lambda
ranges, EtOH base fuel, 1000 RPM, intake temperature of 380 K ................................ 58
Figure 20: Pressure vs. CAD with varying EGR rate for each of 4 lambda ranges, EtOH
fuel, 1000 RPM, intake temperature of 380 K ............................................................... 60
Figure 21: Lifetime of 50 µm ethanol droplets during the compression stroke of Isuzu
4HK1-TC test engine, intake temperature is 370 K, ...................................................... 63
Figure 22: Multi-cylinder test apparatus ........................................................................ 65
Figure 23: Detail of EGR Loop ...................................................................................... 68
Figure 24: Schematic of intake heating bench test ......................................................... 70
Figure 25: Total concentration of particles between 2.5 and 80 nm at heater exit......... 71
Figure 26: Optimization of engine output with intake temperature, ethanol HCCI,
constant fueling, 3 loads, 1500 RPM.............................................................................. 77
Figure 27: In-cylinder pressure behavior of ethanol HCCI combustion, fixed fueling, λ
=5.0-4.2, 1500 RPM, varying intake temperature .......................................................... 79
Figure 28: In-cylinder pressure behavior of ethanol HCCI combustion, fixed fueling, λ
=4.0-3.5, 1500 RPM, varying intake temperature .......................................................... 79
Figure 29: In-cylinder pressure behavior of ethanol HCCI combustion, fixed fueling, λ
=3.2-3.0, 1500 RPM, varying intake temperature .......................................................... 80
Figure 30: Response of combustion and cycle efficiencies to variations in intake
temperature, ethanol HCCI combustion, 3 loads, 1500 RPM ........................................ 82
Figure 31: Brake specific emissions from ethanol HCCI combustion with varying intake
temperature, fixed fueling, λ =5.0-4.2, 1500 RPM ........................................................ 84
vi
Figure 32: Mobility size distributions from ethanol HCCI combustion with varying
intake temperature, fixed fueling, λ =5.0-4.2, 1500 RPM ............................................. 84
Figure 33: Mass distributions from ethanol HCCI combustion with varying intake
temperature, fixed fueling, λ =5.0-4.2, 1500 RPM ........................................................ 85
Figure 34: Brake specific emissions from ethanol HCCI combustion with varying intake
temperature, fixed fueling, λ=4.0-3.5, 1500 RPM ......................................................... 86
Figure 35: Mobility size distributions from ethanol HCCI combustion with varying
intake temperature, fixed fueling, λ =4.0-3.5, 1500 RPM ............................................. 87
Figure 36: Mass distributions from ethanol HCCI combustion with varying intake
temperature, fixed fueling, λ =4.0-3.5, 1500 RPM ........................................................ 87
Figure 37: Brake specific emissions from ethanol HCCI combustion with varying intake
temperature, fixed fueling, λ=3.2-3.0, 1500 RPM ......................................................... 88
Figure 38: Mobility size distributions from ethanol HCCI combustion with varying
intake temperature, fixed fueling, λ =3.2-3.0, 1500 RPM ............................................. 89
Figure 39: Mass distributions from ethanol HCCI combustion with varying intake
temperature, fixed fueling, λ =3.2-3.0, 1500 RPM ........................................................ 89
Figure 40: In-cylinder pressure behavior of ethanol HCCI combustion with varying
EGR rate, fixed fueling, low load, 1500 RPM, 130° intake temperature....................... 96
Figure 41: In-cylinder pressure behavior of ethanol HCCI combustion with varying
EGR rate, fixed fueling, mid load 1, 1500 RPM, 110° intake temperature.................... 97
Figure 42: In-cylinder pressure behavior of ethanol HCCI combustion with varying
EGR rate, fixed fueling, mid load 2, 1500 RPM, 100° target intake temperature ......... 97
Figure 43: Response of combustion and cycle efficiencies to EGR Rate, ethanol HCCI
combustion, 3 loads, 1500 RPM................................................................................... 100
Figure 44: Brake specific emissions from ethanol HCCI combustion with varying EGR
rate, 1500 RPM, 130°C intake temperature, low load.................................................. 101
Figure 45: Mobility size distributions with varying EGR rate, ethanol HCCI
combustion, fixed fueling, 1500 RPM, 130° intake temperature, low load ................. 102
Figure 46: Mass distributions with varying EGR rate, ethanol HCCI combustion, fixed
fueling, 1500 RPM, 130° intake temperature, low load ............................................... 103
Figure 47: Brake specific emissions from ethanol HCCI combustion with varying EGR
rate, 1500 RPM, 110°C intake temperature, mid load 1 .............................................. 103
Figure 48: Mobility size distributions with varying EGR rate, ethanol HCCI
combustion, fixed fueling, 1500 RPM, 110° C intake temperature, mid load 1 .......... 104
Figure 49: Mass distributions with varying EGR rate, ethanol HCCI combustion, fixed
fueling, 1500 RPM, 110° C intake temperature, mid load 1 ........................................ 104
Figure 50: Brake specific emissions from ethanol HCCI combustion with varying EGR
rate, 1500 RPM, 100°C intake temperature*, mid load 2 ............................................ 106
Figure 51: Mobility size distributions with varying EGR rate, ethanol HCCI
combustion, fixed fueling, 1500 RPM, 100°C intake temperature*, mid load 2 ......... 106
Figure 52: Mass distributions with varying EGR rate, ethanol HCCI combustion, fixed
fueling, 1500 RPM, 100°C intake temperature*, mid load 2 ....................................... 107
Figure 53: In-cylinder pressure behavior of EtOH and H2 HCCI combustion, varying H2
output power, 1500 RPM, low load, 130° intake temperature ..................................... 114
vii
Figure 54: In-cylinder pressure behavior of EtOH and H2 HCCI combustion, varying H2
output power, 1500 RPM, mid load 1, 110° intake temperature.................................. 114
Figure 55: In-cylinder pressure behavior EtOH and H2 HCCI combustion, varying H2
output power, 1500 RPM, mid load 2, 95° intake temperature.................................... 115
Figure 56: Response of combustion and cycle efficiencies to variations in H2:EtOH
proportion, dual fuel HCCI combustion, 3 loads, 1500 RPM ...................................... 117
Figure 57: Brake specific emissions from EtOH and H2 HCCI combustion with varying
H2 energy, 1500 RPM, low load, 130°C intake temperature........................................ 119
Figure 58: Mobility size distributions from EtOH and H2 HCCI combustion with
varying H2 energy, 1500 RPM, low load, 130°C intake temperature .......................... 119
Figure 59: Mass distributions from EtOH and H2 HCCI combustion with varying H2
energy, 1500 RPM, low load, 130°C intake temperature............................................. 120
Figure 60: Brake specific emissions from EtOH and H2 HCCI combustion with varying
H2 energy, 1500 RPM, mid load 1, 110°C intake temperature .................................... 121
Figure 61: Mobility size distributions from EtOH and H2 HCCI combustion with
varying H2 energy, 1500 RPM, mid load 1, 110°C intake temperature ....................... 122
Figure 62: Mass distributions from EtOH and H2 HCCI combustion with varying H2
energy, 1500 RPM, mid load 1, 110°C intake temperature.......................................... 122
Figure 63: Brake specific emissions from EtOH and H2 HCCI combustion with varying
H2 energy, 1500 RPM, mid load 2, 95°C intake temperature ...................................... 123
Figure 64: Mobility size distributions from EtOH and H2 HCCI combustion with
varying H2 energy, 1500 RPM, mid load 2, 95°C intake temperature ......................... 124
Figure 65: Mass distributions from EtOH and H2 HCCI combustion with varying H2
energy, 1500 RPM, mid load 2, 95°C intake temperature............................................ 124
Figure 66: Ethanol fueling rate specific CO and HC emissions normalized with respect
to 0% hydrogen fueling ................................................................................................ 125
Figure 67: Brake specific emissions vs. peak in-cylinder temperature, ethanol HCCI
with 0 to 25% supplemental hydrogen fueling, 1500 RPM, 3 loads............................ 127
Figure 68: Brake specific emissions vs. peak heat release rate, ethanol HCCI with 0 to
25% supplemental hydrogen fueling, 1500 RPM, 3 loads ........................................... 127
Figure 69: In-cylinder pressure traces of hydrogen HCCI combustion, fixed fueling, λ =
5.08 - 4.97, 1500 RPM, varying intake temperature .................................................... 131
Figure 70: Effect of intake temperature on combustion and cycle efficiency, hydrogen
HCCI combustion, fixed fueling, λ = 5.08 - 4.97, 1500 RPM ..................................... 133
Figure 71: Brake specific emissions from hydrogen HCCI with varying intake
temperature, fixed fueling, λ = 5.08 - 4.97, 1500 RPM ............................................... 135
Figure 72: Mobility size distributions from a hydrogen fueled HCCI engine,............. 136
Figure 73: Mass distributions from a hydrogen fueled HCCI engine, ......................... 136
Figure 74: BSPM vs. peak HRR or peak temperature, neat hydrogen HCCI, 1500 RPM,
low load, 3 intake temperatures.................................................................................... 137
Figure 75: Neat ethanol and neat hydrogen mass distributions, HCCI combustion, 1500
RPM, Load ≈ 54 Nm, IMEP ≈ 230 kPa, λEtOH = 4.4, λH2 = 5.0 ................................... 139
Figure 76: Full distribution and TDMA data, motoring load, 1500 RPM ................... 145
Figure 77: Full distribution and TDMA data, low load, 1500 RPM ............................ 145
Figure 78: Full distribution and TMDA data, mid load 1, 1500 RPM......................... 146
viii
Figure 79: Full distribution and TMDA data, mid load 2, 1500 RPM......................... 146
Figure 80: Evaporation profiles particulate matter from an ethanol fueled HCCI engine
at three fired loads and a motored load, 1500 RPM ..................................................... 148
Figure 81: Remaining volume fraction of PM in ethanol HCCI exhaust after thermal
conditioning during TDMA analysis, 4 loads, 1500RPM............................................ 149
Figure 82: Fuel and air charge, piston, and cylinder liner interface............................. 150
Figure 83: Particle size distributions collected with and without a catalytic stripper,
motoring and fired engine loads ................................................................................... 153
Figure 84: Average emissions data collected via FTIR spectroscopy from ethanol fueled
HCCI combustion, 4 loads, 1500 RPM ........................................................................ 160
Figure 85: BSNOX v. peak cylinder temperatures for ethanol and hydrogen HCCI with
SOC controlled by multiple strategies.......................................................................... 167
Figure 86:BSCO v. peak cylinder temperatures for ethanol and hydrogen HCCI with
SOC controlled by multiple strategies.......................................................................... 168
Figure 87: BSHC of BSH2 v. peak cylinder temperatures for ethanol and hydrogen
HCCI with SOC controlled by multiple strategies ....................................................... 168
Figure 88: BSPM v. peak HRR for ethanol and hydrogen HCCI with SOC controlled by
multiple strategies......................................................................................................... 169
ix
Nomenclature
BP ......................................................Brake Power
τ..........................................................Brake Torque
k .........................................................Number of Cylinders
N ........................................................Engine Speed, Revolutions Per Minute
Vd .......................................................Volumetric Cylinder Displacement
ηv ........................................................Volumetric Efficiency
F.........................................................Fuel to Air Ratio
hc ........................................................Lower Heating Value
ηc ........................................................Cycle Efficiency
ηm .......................................................Mechanical Efficiency
ηTh ......................................................Thermal Efficiency
ηCombust ...............................................Combustion Efficiency
SI........................................................Spark Ignition
CI .......................................................Compression Ignition
IC .......................................................Internal Combustion
HCCI..................................................Homogeneous Charge Compression Ignition
CAI ....................................................Controlled Auto Ignition
PFI .....................................................Port Fuel Injection
DI.......................................................Direct Injection
EtOH..................................................Ethanol
NOX ...................................................Oxides of Nitrogen
CO......................................................Carbon Monoxide
x
CO2 ....................................................Carbon Dioxide
PM .....................................................Particulate Matter
HC......................................................Unburned Hydrocarbons
PM10 .................................................Particulate Matter <10 µm in diameter
PM2.5 ................................................Particulate Matter <2.5 µm in diameter
NDIR .................................................Non-Dispersive Infrared
CLD ...................................................Chemiluminescence
FID.....................................................Flame Ionization Detector
FTIR ..................................................Fourier Transform Infrared
TDC ...................................................Top Dead Center
BDC ...................................................Bottom Dead Center
SOC ...................................................Start of Combustion
CAD...................................................Crank Angle Degrees
CA10..................................................10 % Cumulative Heat Release Location
CA90..................................................90 % Cumulative Heat Release Location
MFB50...............................................50 % Cumulative Heat Release Location
HRR...................................................Heat Release Rate
EGR ...................................................Exhaust Gas Recirculation
CR......................................................Compression Ratio
CN......................................................Cetane Number
λ .........................................................Excess Air Ratio
Φ ........................................................Equivalence Ratio
γ .........................................................Ratio of Specific Heats
xi
IVO ....................................................Intake Valve Open
IVC ....................................................Intake Valve Close
EVO...................................................Exhaust Valve Open
EVC ...................................................Exhaust Valve Close
NVO ..................................................Negative Valve Overlap
RON...................................................Research Octane Number
MON..................................................Motor Octane Number
IMEP..................................................Indicated Mean Effective Pressure
BMEP ................................................Brake Mean Effective Pressure
NMEP ................................................Net Mean Effective Pressure
xii
Chapter 1
Introduction
Homogeneous Charge Compression Ignition (HCCI) has emerged as a key
technology for the future of the internal combustion (IC) engine. It represents an
evolutionary step in energy conversion as the classic Diesel and Otto cycles merge and
the distinct advantages of both cycles are realized. However, differing from both of
these traditional engine cycles, HCCI engines lack a physical event which controls the
start of combustion (SOC). Depending only on the thermal history and chemical
behavior of the cylinder contents, SOC is manipulated by precise manipulation of these
variables through methods such as; intake temperature conditioning, fuel blending,
exhaust gas recirculation (EGR), variable cylinder geometry, and variable valve
timings. Issues related to energy conversion are broad, sweeping, and leave absolutely
no demographic unaffected. In the simplest of terms we as an industrialized society
compete for the inputs to live and prosper, and in consequence, must deal with the
byproducts of our existence. When we draw analogies in the area of energy conversion
and power generation, the questions to be asked are; what goes in to our engines, how
efficiently do we convert it to useful work, and what comes out? The HCCI engine has
been shown to exhibit flexibility in terms of fuel input, efficiently convert those fuels to
useful work, and generate minimal emissions in comparison with current engine
technology.
1.1 Motivation
HCCI has seen renewed interest from the research community in recent years.
More advanced engine control, volatile fuel prices, and stricter emissions regulations
have motivated researchers to put increased resources into this relatively young engine
technology. Advanced engine control is allowing the combustion mode, once thought to
be impractical for highly variable on-road conditions, to be commercialized in
applications varying in size from small motorcycles to heavy duty industrial Diesels.
Forms of HCCI have been explored by Honda (Ishibashi, 2000), Nissan (Kimura et al.,
1999), and Toyota (Hasegawa and Yanagihara, 2003) in production engines. Both
1
Nissan and Toyota have recently employed strategies that use multiple fuel injections in
Diesel engines at altered timings to enhance mixing thus simulating a premixed burn.
Honda has taken an approach much closer to that of Onishi et al. (1979) utilizing the
natural EGR present in two stroke engines. HCCI engines are well suited to running at
an optimized constant load and speed; with the growing popularity of hybrid vehicles in
consumer markets, HCCI technologies could find their way to mass production on a
series hybrid within 5 years. With any new technology, it is important that we develop a
comprehensive understanding of the mechanisms governing operation and the
consequences of manipulation. The clearly understood benefits of HCCI engines are the
near Diesel efficiencies achieved with simultaneous mitigation of the particulate matter
(PM) and oxides of nitrogen (NOX) problems that have historically plagued Diesel
engines. As pointed out by Price et al. (2007), PM emissions in HCCI engines are often
regarded as negligible. However, few researchers have begun to look in detail at these
particulate emissions. The limited work that has been put forth shows evidence of total
PM mass being drastically cut, while the total number of particles below 50 nm, or
nanoparticles, is observed to increase significantly (Price et al., 2007, Kaiser et al.,
2002, Misztal et al., 2009a, Misztal et al., 2009b). These works show limited cases, but
they do indicate significant need for more thorough study.
1.2 Statement of Problem
The key to applying HCCI technology is control of the onset of combustion
without the aid of a physical event. In Diesel engines, this event is the injection of fuel,
in SI engines, the firing of a spark plug. An HCCI engine must draw in fuel and air, and
then subject it to such conditions that the mixture auto ignites via compression with the
appropriate timing. Various control strategies exist to manipulate the thermal and
chemical conditions of the fuel and air charge. A comprehensive examination of the
effect of these control strategies on emissions at various operating conditions is lacking.
1.3 Significance
The work presented within this dissertation examines the interactions between
common control strategies for HCCI combustion and the effects of these strategies on
2
emissions. More generally, the body of knowledge gained from the study of gas phase
and PM emissions in spark ignition (SI) and compression ignition (CI) engines will be
applied to a new type to internal combustion engine. The end goal of the work is to
identify optimal control strategies for a variety of conditions in terms of emissions. If
HCCI engines are to move into consumer markets they will undoubtedly utilize multiple
combinations of the above control strategies. Understanding the effects of different
control mechanisms, chemical or thermal, will give researchers and designers a valuable
input in developing an optimized control map for a given engine.
1.4 Organization
This dissertation is organized in the following manner. Initially an IC engine
background is given in order to gain familiarity with classic reciprocating engines, the
thermodynamics governing them, and the defining characteristics of each type. An
introductory explanation of HCCI engines is then given in an attempt to relate HCCI
engines to classic IC engines, highlight the benefits, and examine the problems
associated with them. The history of HCCI engines is also presented along with a
thorough examination of the current literature which aids in defining the state of the art
and illustrating how the technology has evolved up to this point.
Because the thrust of the work is concerned with examining emissions, specifically
PM below 50 nm, an overview of IC engine emissions will also be presented. This is
followed by an explanation of the instrumentation used for evaluating PM emissions
throughout this work.
Preliminary HCCI modeling has been conducted. These results will be presented,
with discussion focused on how they relate to the experimental work executed and the
current literature. Preliminary modeling was used to shape the design of the
experimental apparatus, which also be presented.
A series of experiments were executed to explore emissions consequences of
various start of combustion control strategies and develop and understanding of the
formation of emissions in HCCI engines. The results of the experimental work will be
presented and discussed. Connections will be made to preliminary modeling and the
3
current literature. Through discussion of the experimental work, an argument will be
developed with the intent of explaining the origins and behavior of particulate matter in
HCCI engines.
4
Chapter 2
Background
HCCI combustion is a combustion mode with characteristics resembling both spark
ignition (SI) and Diesel or compression ignition (CI) processes. Through a hybrid cycle,
the high efficiency of Diesel engines can be obtained with relatively low levels of the
PM and NOX emissions known to plague them. This is made possible through fully
premixed and very lean (λ>1 or Φ<1) combustion that maintains comparatively low
temperatures throughout the process.
2.1 Engine Fundamentals
In order to more easily relate HCCI combustion to conventional reciprocating IC
engine cycles, a brief review of engine fundamentals is presented. Conventional IC
engines can be split into two groups, each utilizing different gas power cycles to
generate power and do work. The first group, SI engines, follows the ideal Otto cycle
which is characterized by the following four processes: adiabatic and reversible
compression (1-2), constant volume heat addition (2-3), adiabatic and reversible
expansion (3-4), and constant volume heat rejection (4-1). The ideal process is
illustrated in Figure 1.
SI engines described by the ideal Otto cycle utilize a homogeneous mixture of fuel
and air in practice. In the actual Otto cycle, the near stoichiometric fuel and air mixture
is compressed from state 1 to state 2. Addition of fuel to the inducted air has historically
been accomplished by means of a carburetor, with modern SI engines using a fuel
injection system to introduce fuel into the intake manifold. For a very general case,
compression ratios are on the order of 10:1, with pre-ignition cylinder pressures of
roughly 700 kPa and peak pressures near 2000 kPa (Heywood, 1988). Heat addition
takes place from state 2 to state 3 via constant volume heat addition in the form of an
exothermic combustion reaction, which is ignited via electric discharge from a spark
plug. Ignition timing is controlled exclusively by the spark timing, with any type autoignition viewed as undesirable. Moving along the isentrope from state three to state four
is the power generation portion of the cycle where the working fluid undergoes
5
expansion after burning. Finally heat is rejected from the cylinder between states four
and one.
p
3
2
4
1
v
Figure 1: p-v diagram of the ideal Otto cycle
The second group generates power and does work via the ideal Diesel cycle
through the following similar processes: adiabatic and reversible compression (1-2),
constant pressure heat addition (2-3), adiabatic and reversible expansion (3-4), and
constant volume heat rejection (4-1). Figure 2 illustrates the ideal Diesel process.
p
2
3
4
1
v
Figure 2: p-v diagram of ideal the Diesel cycle
6
In the case of a CI engine following the Diesel cycle, much of the p-v path is the
same. The compression ratios are higher however, on the order of 20:1 giving preignition cylinder pressures near 5000 kPa and peak pressures near 15000 kPa
(Heywood, 1988). In contrast to the Otto cycle, the Diesel cycle gains heat via constant
pressure heat addition from state two to state three. This means that the piston is moving
downward and expanding as the mixture burns. The remainder of the piston travel is
accomplished between state three and state four as the hot combustion gas expands.
Similar to the Otto cycle the Diesel cycle is concluded with constant volume heat
rejection in moving from state three to state four. The physical event controlling the
start of combustion in a CI engine is the injection of the fuel into the hot compressed air
within the cylinder. Because no fuel is present within the cylinder until just prior to the
combustion event, much high compression ratios are possible in CI engines.
The most basic analysis of reciprocating IC engines requires consideration of the
entire engine as a control volume, with mass crossing the boundary as air, fuel, and
exhaust gas, work leaving the system as shaft power, and waste heat also leaving the
system. With great simplifications made by considering heat rejection from exhaust gas
and waste heat via efficiencies, analysis of fuel and air inputs and their relationship to
work output is done through Equation 1. In both groups of engines the thermodynamic
cycle must be coupled with the physical processes of introducing charge into the
cylinder and expelling exhaust gas from the cylinder. The entire event may be
accomplished by either two (2-stroke cycle) or four (4-stroke cycle) strokes of the
piston. Air enters the engine and is quantified as a function of ρair, k, N, Vd, and ηv,
which respectively represent; air density, the number of cylinders, the displacement of a
single cylinder, and the volumetric efficiency of the engine at that condition. Fuel
entering is quantified via F, the ratio of mass of fuel to mass of air, and hc, the lower
heating value of the fuel. And work output is given as brake power (BP) through
mechanical and cycle efficiencies. This equation is used in its presented form for a 4
stroke cycle engine.
7
BP = ρ air k
N
Vdη v Fhcη cη m
2
1
Terms commonly used to describe the relationship of actual operating chemistry to
stoichiometric chemistry for a given fuel are λ and Φ, they are defined below by the
mass based ratio between fuel and air, F.
 mFuel 


1  m Air  Stoich
λ= =
Φ  mFuel 


 m Air  Actual
2
Internal combustion engines vary greatly in displacement and total power output.
Mean effective pressure (MEP) has been established as a normalized measure of engine
output and is related to engine power through Equation 3.
Power = MEP × V d × k ×
N
2
3
Either brake power or indicated power can be used in the equation, the
corresponding MEP is then either referred to as BMEP or IMEP. Stone (1999) defines
IMEP directly as indicated work over the displaced cylinder volume. Indicated
parameters and brake parameters are related through the mechanical efficiency of the
engine. Because the only work done by the piston is defined by the integration of
pressure over a changing volume, IMEP can be calculated directly from in-cylinder
pressures via Equation 4.
8
IMEP =
∫ PdV
Vd
4
For the work presented here IMEP will be used to represent engine load. This gives
the ability to quantify results with a normalized parameter that is valid for a variety of
engine sizes.
Heat release rate HRR is used in quantifying in-cylinder engine phenomena such as
SOC, cylinder temperature, and burn duration. Heat release, calculated through the first
law of thermodynamics, is the amount of heat energy added to the cylinder contents to
produce the measured variation in in-cylinder pressure. Neglecting heat transfer to the
cylinder walls, heat release is a function of the internal energy change of the system and
the work (PdV) done by the piston. In evaluating internal energy changes, the cylinder
contents are assumed to be ideal gases. Quantitatively defined through heat release,
SOC is the point at which 10% of the total energy is released, burn duration is then
defined as the time between 10% fuel energy release and 90% fuel energy release.
2.2 Homogeneous Charge Compression Ignition
HCCI is a next generation engine technology capable of exploiting the primary
advantages of both compression ignition and spark ignition engines. The physical
process will first be described then be related to classic IC engine thermodynamic
cycles. As with both a CI and SI reciprocating engines, the reciprocating HCCI engine
is a piston-cylinder device. In four-stroke HCCI the valve train operates in a similar
manner as a common four stroke engine. The piston positions and valve events are
shown in Figure 3. Stroke 1 consists of the piston moving downward and a mixture of
fresh charge, air and fuel, being introduced via the intake valve. The intake valve is then
closed and the mixture compressed as the piston approaches top dead center (TDC),
shown in position two. As the piston approaches TDC the mixture nears its autoignition temperature due to compression heating. Upon reaching its auto-ignition
temperature, the mixture will auto-ignite in a series of spontaneous reactions occurring
9
nearly simultaneously across the cylinder. The device behaves like a homogeneous
reactor ignited via compression ignition. Differing from surface pre-ignition there is no
traveling flame front creating regions of hot, high pressure burned gas. Additionally, the
process avoids uncontrolled knock though highly dilute in-cylinder conditions and
much lower fuel to air ratios than those found in traditional SI engines. This limits high
rates of heat release responsible for damage in SI knocking. The ignition is controlled
solely by chemical kinetics, thus governed by species concentration and mixture
temperatures (Najt and Foster, 1983).
& Air , m
& Fuel
m
Stroke 1
Stroke 2
Stroke 3
Stroke 4
& Exhaust
m
Figure 3: Charge path in a 4 stroke HCCI engine
The third stroke of the cycle represents power generation from expansion of the gas
due to heat addition from combustion. After the piston has reached bottom dead center
(BDC) the exhaust valve opens and the piston moves up in the fourth and final stroke of
the cycle, which is responsible for exhausting the products of combustion.
From a gas power cycle point of view, the HCCI cycle is a variant of the classic
engine cycles exhibiting characteristics of both the Otto and Diesel cycles. Like both
these cycles, the ideal HCCI cycle begins with adiabatic and reversible compression
during stroke one, shown by the isentrope from state 1 to state 2 in Figure 4.
10
p
3
2
4
1
v
Figure 4: p-v diagram of the ideal HCCI cycle
HCCI engines often utilize high compression ratios, similar to those of classic CI
engines, along with constant volume heat addition, generally associated with SI
combustion. In order to attain such high compression ratios with a premixed fuel and air
charge, fuel to air ratios must be very lean or highly dilute, with values of λ on the order
of 3 to 5 not uncommon.
An understanding of HCCI combustion can be achieved relatively simply. Ignoring
for the present the fuel behavior and idealizing the process as polytropic, but still
adiabatic and reversible, we can use the pressure volume relation for a polytropic gas to
relate change in volume, pressure, and temperature (Moran and Shapiro, 2000).
p1V1γ = p 2V2γ
Substituting the compression ratio (CR) for V1 divided by V2 we get Equation 6.
11
5
p2
= CR γ
p1
6
If we add the assumption that the working fluid behaves as an ideal gas, the temperature
at state 2 becomes a function of the state 1 temperature and the amount the gas is
compressed (CR).
T2 = T1CR γ −1
7
This analysis illustrates a key concept in HCCI timing control. The fuel and air
charge must be elevated above its auto-ignition temperature in order for the mixture to
begin burning. Ideal SOC timing occurs at TDC with burn duration kept as short as
possible to approximate the constant volume heat addition of the ideal Otto cycle.
Controlling T2 is accomplished via manipulation of either engine intake temperature or
cylinder compression ratio, with intake temperature being the input most easily varied.
It is now pertinent to address the role of charge chemistry HCCI timing. Autoignition characteristics, unique properties of fuels, are viewed differently in different
types of engines. Auto-ignition is spontaneous ignition of the charge due to its own
thermal energy, which is provided by compression from both the piston and rapidly
expanding burned gas, rather than consumption of the fuel and air mixture by a
traveling flame front. As temperatures increase, reaction rates of chain propagating and
branching mechanisms also increase, creating an uncontrolled combustion event known
as knock. Auto-ignition is undesirable in SI engines and resistance to it in SI fuels is
characterized by the octane number. Octane number is quantified via the research
octane number (RON), motor octane number (MON) or an average of the two (ON).
With common fuels ranging from 60 to 130 or higher, octane number describes a fuels
resistance to knock for a given set of conditions with the highest octane numbers
corresponding to the most knock resistant fuels. Analogously CI engine fuels are rated
12
through cetane numbers with common fuels ranging between 20 and 100, the upper end
of which defines very ignitable fuels. Cetane numbers (CN) characterize a particular
fuels ability to auto-ignite by gauging how readily the fuel auto-ignites after a critical
temperature and pressure is reached within the cylinder. Cetane numbers provide a
comparative evaluation of the ignition delay, or time between fuel injection and SOC,
for a given fuel. The relationship exhibited between cetane and octane numbers is
inverse. Because numbers are found empirically the correlation is not exact but follows
the general trend shown in Figure 5.
Similar to a knock event in SI engines, HCCI combustion results from auto-ignition
of premixed fuel and air due to compression. However, very lean or highly dilute
mixtures maintain a controlled burn by absorbing some of the thermal energy released
during the reactions.
100
80
60
Octane Number
40
20
0
0
20
40
60
Cetane Number
80
Figure 5: Cetane number and octane number relationship (Stone, 1999)
13
2.2.1
Historical Perspective
Originally given the name Active Thermo-Atmosphere Combustion (ATAC) by
Onishi et al. (1979), the HCCI process was developed as a means to reduce emissions
and increase fuel efficiency in 2-stroke SI engines at part throttle loads. The authors
sought to exploit lean burning auto ignition in 2-stroke engines to minimize the cycleto-cycle variability which results in misfire. Misfire leads to high HC emissions in 2stroke engines.
Onishi et al. (1979) contrasted HCCI combustion to standard SI combustion by
examining the manner in which the mass of fuel and air mixture is burned. Rather than
heat release in abrupt and discrete unit mass reactions across the volume of the cylinder,
as in standard combustion, the entire mixture of fuel and air releases its energy in set of
simultaneous reactions. Theoretically, this means that there is no stationary or
propagating flame front, making the process uniquely different from either standard CI
or SI combustion. Although Noguchi et al. (1979) refer to the burning process as flame
propagation; they clearly describe a process of multiple ignition sites and a rapidly
spreading flame in all directions similar to the observations of Onishi et al. Noguchi
also differentiates the process from classic “run on” by presenting Schlieren
photography of combustion initiation at multiple sites away from cylinder walls. In a
situation of run-on, hot carbon deposits ignite the charge, thus flame propagation starts
from the cylinder walls where these deposits are located. Onishi and Noguchi coined
distinct terms ATAC and TS (Toyota-Soken) combustion for their discoveries; however
they were both documenting the same phenomenon. In both cases significant fuel
consumption savings were realized, along with the mitigation of the cycle to cycle
variability normally present in 2 stroke SI engines.
One of the significant and unique characteristics of HCCI combustion is the two
stage heat release exhibited by most fuels (Stanglmaier, 1999). Noguchi et al. referred
to a primary low temperature reaction and secondary combustion reaction along with
identifying the importance radicals from previous cycles in controlling the onset of the
first stage reaction. It was observed that CHO, HO2, and O radicals were present and
behaved as ignition kernels for the initial stage of TS combustion. Upon the formation
14
of significant numbers of OH radicals, the combustion reaction would commence.
Noguchi et al. also hypothesized that the CHO, HO2, and O radicals originated from
chemical cracking of unburned residuals or fresh mixture at the mixing boundary.
HCCI was first researched in a 4 stroke engine by Najt and Foster (1983) under the
name compression ignition homogeneous charge (CIHC). Much of the current work has
moved from the early roots in two stroke research to four stroke work due to the
increased ability to manipulate the details of the gas exchange process. Najt and Foster
(1983) reiterate that there is a lack of a flame front in HCCI engines and further
elaborate on the nature of the 2-stage combustion reactions. They note that the low
cycle to cycle variability seen by Onishi and Noguchi is observed in 4 stroke operation
as well. Most importantly they begin to clearly define which operating parameters are
most pertinent to HCCI combustion and what effects they have. Theoretical and
experimental work clearly showed the advance of combustion timing with increased
intake charge temperature. The role played by EGR was investigated in detail and it was
concluded that few radicals in the exhaust would survive until the next cycle and the
primary effect of EGR on ignition timing is thermal.
The name HCCI was first used by Thring (1989) to summarize earlier efforts by
Onishi et al., Noguchi et al., and Foster and Najt, as well as describe his own work. In
the ten years preceding the work of Thring and the twenty years since, HCCI has been
given at least 10 different titles. The multiple names and acronyms were summarized by
Zhao (2007) in the following statement; “Close examination of these names and the
rationales behind them shows that all names contain the description of two fundamental
characteristics of the new combustion process: 1) premixed fuel and air mixture, and 2)
auto-ignited combustion.” Zhao introduced two names to describe all HCCI activities
up to the time of publication. They differed in that the first term, HCCI, is used to
describe work conducted under a set of circumstances more representative of Diesel
combustion, with low octane/high cetane fuels and relatively high compression ratios.
The second term CAI (Controlled Auto-Ignition) is then used to describe processes with
characteristics showing more similarity to gasoline engine processes, using slightly
lower compression ratios and high octane/low cetane fuels. This distinction was made
15
to illustrate that for high octane/low cetane fuels compression alone is not adequate for
ignition, significant heating of the charge is also required and referring to all process as
simply compression ignition can be misleading (Zhao, 2007). For the work presented
here, HCCI will be used as the name describing the process.
2.2.2
Current Relevant Literature
The current body of knowledge surrounding HCCI engine research has been
evolving for thirty years. Relative to the volume of work addressing conventional IC
engines, there is limited study, however. As a concept HCCI is well researched with
significant amounts of information available from a variety of researchers, however
many gaps are still present concerning a full understanding of the field. Current work
can be broken into a number of subfields, including development of variable engine
geometry, development of HCCI control strategies, chemical and CFD modeling, and
fuels research. Three common goals are shared by most researchers. The first is to
develop a thorough understanding of the HCCI combustion phenomenon. The second is
to use this understanding to exploit a variety of strategies aimed at precisely controlling
the start of ignition. And the third, to further expand the attainable speed and load range
of HCCI engines. The core limitations of this technology are an inability to control
ignition timing and rate of heat release at very low loads and very high loads
respectively. These limitations are manifested in misfire (low loads) and knock (high
loads).
The most common approach for controlling auto-ignition timing in HCCI research
is the use of thermally conditioned intake air. Evolving from early work by Najt and
Foster (1983) and Thring (1989) amongst others, which examined steady state
conditions, current research is focused on fast response thermal management systems
aimed at extending load and speed ranges and transient operation. Realistic sources of
thermal energy for HCCI were outlined by Yang (2005) when investigating its use to
extend the lower load boundaries for high octane (gasoline) fuels. To achieve the
desired intake temperatures while pushing the lower load boundaries, the authors
required utilization of the thermal energy from compression heat, residuals, exhaust
16
gases, and coolant. As an easily varied parameter when compared with compression
ratio or valve timing, thermal management maintains its relevance as a control strategy.
Steady state heating via resistive elements sees extensive use in research applications,
however, they would serve little use under the rapidly changing conditions of real world
engine operation. Understanding that a conventional heating system would have far too
slow of response due to its high thermal inertia, mixing systems comprised of separate
hot and cold streams of intake air have seen significant research and allow rapid
adjustment hot and cold proportions based on engine demands. A thermal management
system recently developed by Martinez-Frias et al. (2002) and further refined by
Flowers et al. (2005) allows for rapid response to changing thermal demands via closed
loop feedback controlled mixing valves. In a 6-cylinder HCCI engine Flowers has
implemented an advanced intake manifold utilizing independently heated and cool air
streams for each cylinder. The closed loop feedback is obtained via combustion timing
calculated from in-cylinder pressure. This adiabatic mixing methodology is similar to
that of Peng et al. (2007) who utilized a set of mixing valves controlling the proportion
of ambient and heated air to a single cylinder engine. Additionally Peng et al.
successfully operated this system through the transition between SI mode and HCCI
mode. Some fluctuations in engine speed and BMEP were reported at the transition,
they lasted only a few engine cycles. The naturally aspirated single cylinder engine used
by Peng et al. was fueled with ethanol via port fuel injection and utilized EGR as well
as intake heating. Haraldsson et al. (2004) have also developed a rapid response thermal
management system. Differing from Flowers et al., Haraldsson et al. developed a
mixing system on a five cylinder engine without individual cylinder mixing capabilities.
The feedback for the thermal management in this case was an average IMEP across the
five cylinders. Cylinder to cylinder variations in IMEP then were mitigated via an
additional algorithm that controlled fueling to each cylinder and subtly adjusted
individual cylinder fueling to maintain a user determined IMEP. Successful
implementation of a thermal management system similar to Peng et al. was reported by
Hyvönen et al. (2004) with quantitative findings shown. The authors achieved stability
across five cylinders with less than 1 CAD variation in CA50 compared to 11 CAD of
17
variation in CA50 with constant intake air temperatures. Comparing their work to a
cylinder by cylinder fuel management strategy, similar to that of Haraldsson et al., the
authors viewed thermal management as a superior methodology. This was justified by
examining NOX variations from cylinder to cylinder. When fueling was used to bias
combustion phasing, the authors reported different loading and pressure rise rates in
individual cylinders. The result was elevated NOX levels in the cylinders with slightly
higher loads.
The second most common approach for SOC control in HCCI engines is modulated
EGR. In applications using 2-stroke engines, such as those of Onishi et al. (1979) and
Noguchi et al. (1979), EGR provided all necessary thermal energy to facilitate
combustion. In 2-stroke applications EGR is present in the form of residuals left in the
cylinder between cycles. In the cases of EGR as residuals it is also possible that radicals
remain in the burned gas and facilitate combustion reactions. This effect was what early
researchers had in mind when coining the term “Active Radical” combustion. When
first transferring the technology to a 4-stroke engine, Najt and Foster (1983) and Thring
(1989) noted the importance of external EGR due to the lack of hot residuals. Thring
also systematically examined EGR as a diluent to eliminate knock. EGR levels, defined
by CO2 volume concentration though Equation 8, were varied from 13 to 33 %. Higher
rates generally were required to achieve higher equivalence ratios, which correspond to
higher loads.
EGR =
CO2, Intake
CO2, Exhaust
× 100%
8
This work clarified the dual purpose nature of EGR in HCCI engines. The first role
of EGR is raising cylinder temperatures through transfer of thermal energy from hot
exhaust gases to the cooler intake charge. The second role is that of a thermal sink,
controlling the rate of heat release and inhibiting rapid and uncontrollable pressure rise.
Insight into this role can be gained by examining the specific heats of the primary
18
components of exhaust gas and comparing them with the specific heat of air at typical
combustion temperatures. Figure 6 shows the specific heats of the primary components
of exhaust gas and air. Using EES software to model the specific heats, we assume that
air, nitrogen, water, and carbon dioxide are ideal gases at these temperatures. Shown on
a molar basis, the clear difference in the specific heats of CO2 and H2O, when compared
to air, explain the ability of exhaust gases to limit the rate of pressure rise as described
by Thring (1989). Equation 7 is useful clarifying this effect. T1 and T2 are related
through compression ratio (CR) to the exponent γ-1. For ideal gases in an adiabatic and
polytropic process, γ is equal to the ratio specific heats. As this ratio increases, T2 will
exhibit an increase for the same T1.
70
CP,CO2
CP,H2O
CP,N2
CP,Air
60
CP(kJ/kmol·K)
50
40
30
20
10
0
0
500
1000
1500
2000
Temperature (K)
2500
3000
3500
Figure 6: Specific heat (cP) of primary exhaust gas components and air
EGR effects were more recently examined on a multi-cylinder production Diesel
engine modified for HCCI experiments by Au et al. (2001). Experiments were
conducted that maintained constant intake temperature, effectively removing the
thermal role of EGR, and varied overall EGR rate. Burn duration was used to quantify
combustion behavior. Au et al. defined the burn duration as the number of CAD
19
required to complete 10-90% of the total heat release. It was found that total burn
duration increased significantly with EGR rate, but SOC had little or no dependence on
EGR rate.
Lü et al. (2005b) examined in detail the effects of cooled EGR on SOC and
combustion duration. In agreement with Au et al. (2001), the authors saw very little
effect of cooled EGR on SOC. These experiments were done for reference fuel blends
ranging in octane number from RON 0 to RON 75. For n-heptane (RON 0), peak
pressures and total burn duration showed little dependence on EGR as rates. However
higher octane fuels, RON 50 and RON 75, exhibited significant response to increased
EGR rates. In particular with RON 75 fuel, increasing the EGR rate resulted in a shift in
peak pressures later in the cycle by as much as 15 CAD. The results could be observed
for EGR rates ranging from 15% to 45% with the highest rates corresponding to the
greatest CAD shift in peak pressures. In addition, the magnitude of the peak pressure
was drastically cut as more EGR was added. Peak pressures of RON 75 fuel were cut
nearly in half, from 9 MPa to 5 MPa, when 45% EGR was utilized. The authors cited
both the increased heat capacity and lack of oxygen as reasoning for the delay in, and
reduction of, peak pressures. The lack of oxygen was cited as playing an important role
in limiting the primary high temperature chain branching reaction shown in Equation 9.
H + O2 ⇒ O + OH
9
Additionally Sjöberg et al. (2007) have clearly demonstrated a retarding effect of
EGR on ignition timing. The primary reasons for the effect are listed as; first, the high
specific heat of the gases reduces the compressed gas temperature, and second,
reductions in oxygen concentrations limit available oxygen for combustion reactions.
The sensitivities of HCCI combustion to each of these drivers were also documented by
Sjöberg et al. (2007) to vary with fuel type.
The third area of HCCI research most pertinent to this work is utilization of fuel
properties to control combustion. Many types of fuels are suitable for HCCI combustion
20
and a great deal of research examining them has been done. High octane fuels such
gasoline (Thring, 1989), natural gas (Yap et al., 2004), and hydrogen (Gomes Antunes, 2008) have all seen successful application. High cetane fuels such as Diesel
(Tsolakis, 2005) and n-heptane (Lü et al., 2005b) have also been used. Renewable fuels
also show promise, ethanol (Mack, 2009), DME (Shudo, 2003), and bioDiesel
(Tsolakis, 2005) have all been used as primary fuels in HCCI engines. This is not a
comprehensive listing, but rather a sampling to illustrate the flexibility of HCCI
technology for combusting a variety of fuels.
Altering the properties of a single fuel during operation is obviously not feasible;
however modern port fuel injection systems allow blending of two fuels with different
properties on a cycle-to-cycle timescale. Utilization of two fuels allows manipulation of
SOC timing by exploiting the different relative tendencies of different fuels to autoignite. Logistically having two fuels creates problems in terms of on-board storage and
refueling infrastructure. However, work by Deluga et al. (2004) involving autothermal
reforming has shown that modest amounts of hydrogen rich gas can be reformed from
ethanol with relative ease. More recently, a comprehensive look at various types of
reforming bio-ethanol to obtain hydrogen rich gases was given by Ni et al. (2007). Such
small scale reformers could be integrated into an engine fuel system so that hydrogen
could be used as a supplemental fuel, providing on the order of 20% or less of total
energy input. Reforming ethanol to make hydrogen then allows operation of a dual fuel
engine requiring only one type of fuel to be filled and stored. The effectiveness of
producing hydrogen rich gas by reforming conventional hydrocarbons has spurred a
large amount of research in the area of dual fuel engines, particularly dual fuel HCCI
engines. HCCI research utilizing dual fuel systems has seen proof of concept level
research as well detailed combustion analysis by numerous sources.
Yap et al. (2004) presented modeling results that predicted the start of combustion
in an HCCI engine fueled with natural gas and supplemented with varying amounts of
hydrogen. As hydrogen content of the fuel was increased from 0% to 20%, the predicted
start of combustion advanced by nearly 10 CAD. These results suggested hydrogen
addition aided the fuel in igniting more readily so that intake heating requirements
21
could be relaxed while maintaining constant peak pressure timing. Validating the work
experimentally, it was found that the addition of a small amount of hydrogen in the
EGR stream allowed the required intake temperatures to be dropped by as much 20°C at
low loads. Specifically, these exercises examined the temperature drop that was
acceptable with the addition of hydrogen while maintaining a constant indicated mean
effective pressure (IMEP). However, the effect was found to be less profound at higher
loads. Pressure rise rates were then examined and it was found that hydrogen addition
gave slight increases in maximum rates of pressure rise. Again these effects were more
pronounced at low loads. Compression ratios on the experimental engine ranged from
12:1 to 14.5:1, with intake heating on the order of 140° to 300 °C used. Yap et al.
(2006) expanded on the previous work with additional experimental work examining
higher concentrations of hydrogen flow rates in the EGR stream and higher
compression ratios. Again, advancement of the start of combustion was shown with
increasing amounts of hydrogen addition. This was quantified by examining the 5%
burn point, which advanced by roughly 6 CAD at low loads and 3 CAD at high loads.
The amounts of hydrogen used here were very small, on the order of 0.5 % to 0.75% of
total intake air flow, equating to 3 % to 5 % of the total fuel energy. The authors
explained enhancement of combustion via an increase in atomic hydrogen to feed the
chain branching mechanism shown in Equation 9.
Hosseini and Checkel (2006) also investigated the effects of hydrogen rich gas
(reformer gas) on natural gas fueled HCCI combustion. The hydrogen rich gas used in
this work was 75% hydrogen and 25% carbon monoxide. These tests examined
compression ratios ranging from 16:1 to 18.5:1 and intake temperatures of 140°C on a
single cylinder engine. The hydrogen rich gas tended to advance the SOC, agreeing
with the work of Yap. In an unthrottled engine, load is inversely proportional to λ. At λ
values ranging from 2.5 to 2.8, it was shown that using reformer gas to comprise 60%
of the total fuel mass, which is equal to roughly 75% of fuel energy, would advance the
SOC by nearly 6 CAD. Lesser fractions were also evaluated, giving a relatively linear
response. Hosseini and Checkel also found that adding reformer gas to natural gas
HCCI combustion extended the low load end of the operating window, with higher
22
proportions of reformer gas showing increasingly significant effects. In these
experiments EGR rate and compression ratio were held constant and an operating
window was established by varying λ until the engine either lost power or knocked.
Hosseini and Checkel assessed reformer gas effects on both high octane fuels
(2007a) and low octane fuels (2007b). In the high octane tests PRF fuels with octane
numbers of 80 and 100 were tested with compression ratios of 16:1 and 14.4:1
respectively at intake temperatures of 140°. For the low octane tests, n-heptane which
defines ON=0 was used and compression ratios were dropped to 9.5:1 and 11.5:1, with
intake temperatures held at 100°C. These results provide interesting insight into the
difficulty encountered when trying to develop a general fuel index as explained by
Shibata and Urushihara (2007). All of the blends tested were blends of n-heptane and
iso-octane primary reference fuels. In contrast with the above work conducted using
natural gas, the SOC in all of these tests was delayed with increasing amounts of
reformer gas. In the high octane tests with ON=80 fuel, SOC was shifted later by 7
CAD with 30% of the fuel by mass reformer gas when compared to tests with 0%
reformer gas. The same trend was observed in the low octane tests and was even more
pronounced. SOC for ON=0 fuel burned at a constant λ ,EGR rate, CR, and intake
temperature shifted later in the cycle by 12 CAD as reformer gas mass fraction was
increased from 0% to 20%. Interestingly these effects are completely opposite of the
effects seen by Yap (2004, 2006) as well as Hosseini and Checkel (2006) when natural
gas was the primary fuel. The behavior is likely the result of an averaging effect in
octane number when the more easily ignited hydrogen is mixed with natural gas which
has a RON near 120. Hosseini and Checkel (2008) repeated the earlier experiments with
lower levels of hydrogen in the reformer gas, 50% H2 compared to 75% H2 in prior
studies, in order to more realistically simulate actual reformer gas levels of hydrogen.
The results confirmed prior work with natural gas SOC advancing and primary
reference fuel blends retarding as reformer gas mass fraction was increased. The
experimental results were also modeled in ChemComb-SZM and showed similar trends
as the experimental work.
23
2.3 Emissions
Initially examining the emissions characteristics of SI and CI engines gives a set of
base conditions to which comparisons of HCCI emissions can then be made. A
summary of general emissions trends of each of the three types of engines operating at
steady state is given below. In each section a brief description of gas phase emissions
will be presented with greater attention paid to PM emissions.
Engine emission standards regulate a common set of emissions universally agreed
upon as having negative effects on air quality. In the gas phase, CO, HC, and NOX are
regulated, with CO2 not regulated by most legislation but still viewed as undesirable due
to its classification as a greenhouse gas. NOX is primarily composed of NO with only
trace amounts of NO2 in SI engines, however in CI engines NOX can be composed of up
to 30% NO2 (Hilliard and Wheeler, 1979). Engine emissions are commonly normalized
to engine output and presented in the form grams per brake horsepower hour (g/bhp) or
grams per kilowatt hour (g/kW hr).
Solid and liquid phase emissions are also regulated and commonly referred to as
particulate matter (PM). Differing from gas phase emissions, PM emissions are much
more sensitive to sampling conditions and great care must be taken when collecting
these samples. As discussed by Kasper (2005), in modern engines, particles due to gas
to particle conversions in the exhaust system and PM in dilution air can be more
abundant than PM originating from the engine itself. Particulate matter from engines
ranges from less than 5 nm to greater roughly 10000 nm in mobility diameter and, when
examined in the ambient, is distributed trimodally between a coarse, accumulation, and
nucleation modes. The coarse mode contains particles on the order of 1000 to 10,000
nm, the accumulation mode from 30 to 700 nm, and the nucleation mode particles are
generally less than 30 nm (Kittelson, 1998).
The two driving forces behind HCCI research are gaining higher efficiency and
curtailing regulated emissions. In general CI engines have thermodynamic efficiencies
20% to 30% higher than a comparable output SI engine. This is a benefit of the high
compression ratios and lean burn strategies allow by CI. Additionally CI engines
24
eliminate the throttling losses characteristic of SI engines operating at part load. It is
clear from the most elementary combustion chemistry reaction that an increase in
efficiency translates directly to a decrease in CO2 per unit power output.
2.3.1
Spark Ignition Emissions
SI engines power much of the passenger fleet throughout the U.S. Recent work
examining PM source apportionment by Johnson et al. (2005) reports roughly 90% of
the traffic flow on a typical urban interstate to be light duty SI powered vehicles on
weekdays with that number climbing to 99% on weekends. Operating generally very
close to a stoichiometric air to fuel ratio (λ=1), deviation from these conditions has
negative consequences on emissions characteristics. As charge mixtures get
progressively more fuel rich, HC and CO emissions tend to increase, while NOX
emissions peak just lean of stoichiometric and drop off with increasing or decreasing λ.
HCs originate in the crevice volumes around piston rings and in the cool boundary layer
of oil at the cylinder walls. HC emissions will also increase sharply at very lean
conditions as misfire begins to occur (Sher, 1998). CO and NO are both formed in the
high temperature, high pressure burned gas behind the flame front (Mattavi and Amann,
1980). As the combustion chamber volume expands and the fuel is consumed, these
burned gases cool abruptly, freezing molar concentrations of the radicals. Control of
NO concentrations is achieved in part via EGR in SI engines. EGR rates ranging from
10-25% of total intake air are common before combustion becomes unstable. EGR
reduces cylinder temperatures in SI engines, therefore curbing NO production rates, by
acting as a diluent and soaking up thermal energy (Abd-Alla, 2002). In addition to
EGR, implementation of 3-way catalysts in SI exhaust systems is a useful tool in
controlling NOX, CO, and HC emissions. However these catalysts require tightly
controlled combustion conditions very near to stoichiometric to be effective (Twigg and
Wilkins, 1998).
Kayes and Hochgreb (1999a) thoroughly examined PM formation in SI engines
and found experimentally that both total mass and number concentration, as well as
number weighted mean and mode particle size, were at a minimum near stoichiometric.
25
All of these descriptors increased as the mixture moved both lean and rich. In the case
of total mass concentration, increases of 2 orders of magnitude were observed in the
very lean and very rich regions as lambda was swept from .7 to 1.7. The same authors
also reported that as engine load increased both PM total mass and total number
increased as well. In the same study EGR rates were varied while keeping all other
operating parameters fixed. An inverse relationship between EGR rate and total number
and mass concentrations was shown. Examining the composition of PM in SI engines
more closely, Burtscher et al. (1998) found particles to be in the form of chain
agglomerates composed primarily of carbon and a large fraction of volatile material.
Stoichiometric compressed natural gas (CNG) engines have seen limited use as an
alternative to conventional gasoline fueled SI engines in recent years in part due higher
efficiency gained though high compression ratios, and also due to an overall cleaner
burn (Ayar, 2006). Additionally the higher hydrogen to carbon ratio of the fuel creates
less CO2 emissions per unit fuel energy (Cho, 2007). Gas phase CNG encounters no
mixing problems and will not exhibit the fuel rich jet burning characteristics of direct
injection spark ignition (DISI) or CI engines. Additionally, the pooling effects which
can create emissions problems in PFI engines are not present in these types of engines
because of the gas phase fueling strategy. Recent work by Schreiber (2007) comparing
emissions from gasoline fueled SI, Diesel fueled CI, and CNG fueled SI engines
showed the CNG fueled engines producing total particle numbers on par with
concentrations from a Diesel engine found downstream of the DPF. Both of which were
lower than engine out exhaust concentrations in the port fuel injection (PFI) or direct
injection (DI) gasoline engines tested. The results were average concentrations from 19
gasoline fueled cars, 12 Diesel fueled cars, and 3 CNG fueled cars, and were based on
measurements recorded during the New European Driving Cycle (NEDC). Following
similar trends as gasoline fueled SI engines, Ristovski et al. (2000) showed both CMD
(count median diameter) and total concentration (particles/cm3) increasing directly with
load in natural gas fueled SI engines.
26
2.3.2
Compression Ignition Emissions
The overall efficiency gain in CI engines comes at a price when examining PM
emissions. Furthermore, relatively low exhaust temperatures prohibit the use of
catalysts to treat NOX emissions. The excess air of lean burn combustion also
contributes to NO production by ensuring a sufficient source of oxygen. However, EGR
is used extensively to control NO formation in CI engines. In this case, the reductions in
NO concentrations are primarily a result of the displacement of intake air (O2, N2) with
exhaust gases (CO2, H2O, N2), thus limiting available oxygen for NO formation
(Ladommatos, et al., 1999).
Although the global equivalence ratio is maintained lean in Diesels, the nature of
the fuel injection process leads to localized rich regions around the fuel jet as it
undergoes diffusion burning during the burn process (Dec, 1997). This localized rich
combustion produces the sooty black smoke that comes to mind as one considers a path
traveled behind an old Diesel truck. A distinct tradeoff exists between NOX and PM,
when utilizing EGR in CI engines. The NOX – PM tradeoff is well documented by
numerous authors and summarized by (Ladommatos, et al., 1999), Abd-Alla (2002),
and Zheng et al. (2004). The phenomenon is a consequence of the localized rich regions
of diffusion burning being further starved of oxygen when EGR displaces intake air.
The composition of CI, or Diesel, PM is reported by Kittelson (1998) to be
primarily carbon and unburned hydrocarbons, as well as sulfate, water, and ash.
Burtscher et al. (1998) notes that the volatile fraction composing CI particles is much
lower than that of SI engines. An early study conducted by Tobias et al. (2001) used
thermal desorption particle beam mass spectrometry (TDPBMS) and temperature
programmed thermal desorption (TPTD) to investigate Diesel particulate matter
composition. They conducted analysis of the organic compounds comprising Diesel
particulate matter and found high percentages of alkanes and cycloalkanes. The
similarity of the cycloalkanes to alkanes ratios in the PM to those of the lubricating oil
led the authors to believe that significant contributions were made from lubricating oil.
Sakurai et al. (2003) further investigated Diesel nanoparticle composition through
TDPBMS and tandem differential mobility analyzer (TDMA) techniques and found
27
particles emitted at low to moderate loads were composed of roughly 95% compounds
from unburned lubricating oil.
The agglomerative nature of these particles is shown in detail by Park et al. (2003).
The structure and packing density of the primary particles has been shown to vary
depending on the fuel, engine, and operating conditions. Park et al. (2003) also
documented a size dependant difference in effective particle density through a series of
experiments using an aerosol particle mass analyzer to relate particle mass to mobility.
These authors found a trend of decreasing effective density with increasing particle size
ranging from 1.2 g/cm3 at particle sizes near 50 nm to .3 g/cm3 at particle sizes near 300
nm.
In general, due to the abundance of oxygen in lean burning CI engines, CO
emissions are negligible. Although CI engine fuels are less volatile, or more difficult to
ignite, HC emissions levels are generally lower in CI engine than in SI engines due to
high temperatures and global oxygen rich conditions. They often originate from either
misfire due to lack of fuel, or under mixed fuel and air which create localized rich
regions that are not fully oxidized during combustion (Sher, 1998).
2.3.3
Homogeneous Charge Compression Ignition Emissions
Emissions in HCCI engines show clear advantages over SI and CI engines. The
primary hurdle in utilizing HCCI engines, controlling the start of combustion, is also
closely tied to the emissions benefits realized in these engines. In-cylinder combustion
processes dictate the speed and uniformity in which fuel is oxidized and products are
formed. In SI combustion a traveling flame front is initiated at the spark plug and leaves
behind it a region of high temperature burned gas that is further compressed and heated
as the flame continues across the chamber. As mentioned above CO and NO form in
this burned gas. The same start of combustion event is controlled by fuel introduction in
CI engines. However combustion time scales are much faster than mixing timescales
leading to poorly mixed fuel and air combined with diffusion burning of fuel rich
regions. From this diffusion burning precursors to PM formation are created. HCCI
combustion has no event to initiate combustion; reactions are simultaneous and as a
28
result no burned gas effects are present (Zhao, 2007). Additionally fuel and air are
premixed, minimizing diffusion burning.
The allure of HCCI combustion is rooted in the Diesel like efficiency exhibited by
these types of engines accompanied by ultra low emissions. Resulting from premixed,
low temperature combustion, the NOX – PM tradeoff prevalent in CI engines is avoided
and simultaneous reductions can be achieved. Review papers from Johansson (2007),
Juttu et al. (2007), Epping et al. (2002), and Stanglmaier and Roberts (1999) cite
numerous authors noting significant and simultaneous reduction in PM and NOX.
Although the magnitudes of these reductions vary across engine conditions, fuels, and
HCCI control strategies, a clear and well documented trend is indeed present. Figure 7,
adapted from the work of Kitamura et al. (2002), shows regions of soot and NOX
formation in combustion processes obtained via modeling. It can be seen that low
temperature, lean combustion is the ideal operating regime for minimizing both PM
(soot) and NOX. The SI and HCCI regions are well defined and easily understood in this
figure, however the notion that Diesel flames burn rich when the overall air to fuel
proportions are lean, warrants discussion. As mentioned in section 2.3.2, a fuel rich
region initially surrounds each evaporated fuel packet in CI combustion. This is a result
of evaporation of the fuel upon injection and gives way to rich premixed combustion in
that region which located at the upper left most point on the contour path shown. Zhao
(2007) defines 3 regions of combustion in Diesel engines that occur after the initial
ignition delay, with premixed combustion as the first. As we move down the contour to
roughly Φ=3, the mixture begins to burn as a diffusion flame. Due to turbulent mixing,
the partially oxidized hydrocarbons and new fuel eventually move past the flame front
and are burned at high temperatures as locally stoichiometric conditions are approached.
In this burn region maximum temperatures are reached on the contour. After significant
burning has occurred, the globally lean conditions continue to supply oxygen as the
remaining fuel is burned in an increasingly lean and cool environment until all of the
fuel is consumed. It can be seen that inhomogeneities in the fuel mixture of Diesel
engines lead to production of both soot and NOX, a problem that HCCI specifically
addresses.
29
66
55
44
Φ
33
SI Region
22
11
Typical Diesel
Fuel Combustion
Path
HCCI Region
1000
1400
1800
2200
Temperature (K)
2600
3000
Figure 7: Regions of soot and NOX formation in combustion systems
In beginning the discussion of emissions in HCCI engines it will be helpful to
briefly discuss the routes of formation of common harmful emissions during
combustion. Thermal NOX, formed via the Zeldovich mechanism, is the primary
contributor to NO formation during combustion. At ambient pressures the forward
reaction rates of Equations 10, 11, and 12 all increase exponentially as temperatures
surpass 1600 Cº, giving rise to rapidly increasing NO levels (Zhao, 2007). Just as is the
case in SI and CI engines, some of the NO is further oxidized into NO2 as cylinder
temperatures decrease during the expansion stroke. With the combination of these two
compounds referred to as NOX. HCCI combustion is a type of low temperature
combustion and achieves low NOX levels by maintaining in-cylinder temperatures near
or below the high thermal NOX formation regimes. A strategy intended to minimize
NOX formation in any type of engine must either limit peak temperatures or suppress
the reactants feeding the Zeldovich mechanism. Peak temperatures are limited by the
minimization of peak pressures and rates of pressure rise. The most significant
30
reductions in NO in conventional engines are realized with the use of EGR. In the case
of Diesel engines, reactants (air) are displaced with exhaust gases containing CO2 and
H2O. And in the case of SI engines peak temperatures are controlled by the relatively
high specific heats of the same compounds.
N 2 + O ⇔ NO + N
10
N + O2 ⇔ NO + O
11
N + OH ⇔ NO + H
12
Decreasing cylinder temperatures indefinitely is clearly not the answer to cutting
emissions, at some point a combustion event will simply not occur and we will be
pumping fuel in and fuel out of the combustion chamber resulting in no power output
and obviously high HC emissions. However before this most drastic case occurs,
notable trends in both HC and CO emissions become evident. As peak cylinder
temperatures decrease, increases in HC and CO emissions are encountered. Au et al.
(2001) illustrates a tradeoff between NOX and both CO and HC emissions. As EGR
rates are increased to increase burn duration, NO is observed to fall while CO and HC
increase. Oxidation of CO to CO2 occurs at temperatures above roughly 1200 Cº (Zhao,
2007). The HC oxidation reactions also require relatively high temperatures to progress.
Unburned hydrocarbons from fuel in crevice volumes and from fuel and oil at the
cylinder walls increase as cylinder temperatures reach lower peak values. HCCI
combustion temperatures ideally fall in the tightly defined temperature regime that
allows maximum oxidation of CO and HCs, but is below temperatures that promote
rapidly increasing NO levels. The review papers cited above all note increased CO and
HC emissions as a major drawback of HCCI engines. Lü et al. (2005a) show notable
differences in CO and HC emissions between high octane and low octane fuels. At the
same engine conditions both CO and HC emissions were shown to increase directly
31
with octane number. Regardless of octane number, CO and HC emissions were also
shown to decrease with decreasing λ (increasing load). For this particular study CO
emissions of all fuels tested converged and showed little sensitivity to λ at values below
roughly λ = 3. HC emissions of low octane fuels (RON 0, 25, and 50) also decreased
and converged with decreasing λ. These emissions leveled out around a λ value of 4.
For high octane fuel (RON 75) Lü et al. (2005b) has clearly illustrated the ability
of increasing EGR to delay combustion and significantly cut in-cylinder peak
temperatures. A slight dependence of both CO and HC emissions on EGR rates up 40%
was also shown in this work. However an increase of nearly 4 times in HC and CO
emissions was found as EGR rates were pushed from 40% to 45%. Prior to the drastic
increase, these emissions levels were already high, about .3% by volume CO and .05%
by volume HC. These levels are similar to engine out SI levels and far in excess of
average CI levels. Modern SI engines however employ a 3-way catalyst to significantly
cut both CO and HC emissions as well as NOX. Because exhaust temperatures are lower
than SI exhaust temperatures, current oxidation catalyst technology is not effective in
removing CO and HC from the HCCI exhaust stream (Epping et al., 2002).
Currently there have been few studies conducted that closely investigate PM
emissions in HCCI engines. Kaiser et al. (2002) examined PM emissions of an HCCI
engine fueled with gasoline which used an early direct injection strategy at various
timings. A SMPS and two stage ejector dilutor setup was used for PM measurements.
HCCI combustion was achieved with intake temperatures ranging from 150° to 200°C
at a compression ratio of 15.2:1 and λ ranging from 2 to about 18. Particle size
distributions are presented for 3 HCCI operating conditions (λ=2.35, 3.25, and 6.77) at
a single engine speed of 1100 rpm. In addition, particle size distributions from the same
engine running in a DISI (λ=1) mode and a motored mode are shown. At two of the
HCCI conditions, λ=2.35 and λ=3.25, accumulation mode number concentrations were
higher than that of DISI operation. Additionally the mode itself was found at a larger
mobility diameter. The presence of the large accumulation modes was explained by the
existence of at least some degree of diffusion burning. This was a consequence of the
32
DI nature of the fuel injection system and the cool operation of lean burning HCCI
combustion which creates the need for longer evaporation and mixing times to fully
vaporize and mix the fuel. The λ=6.77 HCCI condition showed far fewer accumulation
mode particles than either of the above mentioned HCCI conditions or the DISI
condition. However an order of magnitude increase was seen in nucleation mode
number concentrations at this condition when compared with the other conditions. The
authors also presented HC emissions data at each condition noted that the HC emissions
at the λ=6.77 condition were an order of magnitude higher that the other HCCI
conditions. With extremely lean operation, cylinder temperatures are much cooler,
impeding full oxidation of boundary layer and crevice bound hydrocarbons, thus
increasing concentrations of hydrocarbon precursors to nucleation. Furthermore, the
lower fueling rate at this condition corresponds to shorter injection times. This
represents an increase in mixing time which serves to created a more homogeneous
mixture and minimize diffusion burning. However, the reduction of accumulation mode
particles may have facilitated nucleation of new particulate matter resulting from a lack
of adsorption and condensation sites Kittelson et al. (2003).
The authors report a sharp change in the CO/CO2 proportion as a function of λ,
with CO2 most prevalent at λ<4.5 and CO becoming most prevalent at λ>6.9. As
combustion takes place at leaner conditions, in cylinder temperatures generally decrease
due to charge dilution, impeding full oxidation of CO to CO2. Increased hydrocarbon
emissions are also seen as a consequence of lower in-cylinder temperatures. Similar
results were also reported from a CHEMKIN model by Dec and Sjöberg (2003) and
verified experimentally through HCCI engine tests fueled by iso-octane.
A study by Price et al. (2007) reported PM emissions again from a gasoline fueled
HCCI engine. In this work, a multiple electrometer based differential mobility particle
sizer (Cambustion DMS500) was used to collect PM data. A DI-HCCI fueling strategy
very similar to the one of Kaiser et al. (2002) was also used. Valve timings with
negative valve overlap (NVO), referring to the exhaust valve being closed early at the
end of the exhaust stroke, were used to trap residuals and add thermal energy to the
charge. A single λ of 1 was used for most testing, with the valve timings and intake
33
temperature varied and the emissions reported as a function of these parameters. For all
of the conditions, composed of varied valve timings and intake temperatures, a bimodal
number distribution is reported. A total of 19 different valve timing combinations were
examined and each showed a nucleation mode that had a significantly higher particle
concentration than the accumulation mode of the same distribution. Comparing one
HCCI data set with a SI data set at the same indicated mean effective pressure (IMEP),
the authors found a notable increase in accumulation mode particles and decrease in
nucleation mode particles in the HCCI data compared with the SI data. These results are
in agreement with Kaiser et al. (2002) where the authors noted some degree of diffusion
burning is present due to the DI nature of this type of HCCI. The higher cylinder
temperatures associated with a propagating flame front aid initial droplet evaporation
for the SI cases, giving less diffusion burning and consequently a smaller accumulation
mode in the PM distributions.
A more detailed look at HCCI particle size distributions was recently published by
Misztal et al. (2009a). Similar to the two previous studies, the fuel injection strategy
used was a DI-HCCI system injecting unleaded gasoline directly into the cylinder
employing NVO to capture residuals for thermal energy. The authors utilized an
electrometer based aerosol measuring system (Cambustion DMS500) preceded by
exhaust dilution. However they note that dilution air temperature and humidity were not
tightly controlled. Works by Abdul-Khalek et al. (1999, 2000), Mathis et al. (2004), and
Rönkkö et al. (2006) all illustrate the sensitivity of nanoparticle formation to dilution
temperature and humidity.
The primary focus of this work was to explain the consequences of intake heating
on PM emissions. A mixing system for hot and cold intake streams was designed to
vary intake temperature. In order to examine multiple intake heating temperatures and
different valve timings, a fixed IMEP and engine speed was used for a given case. IVO
was then changed and intake temperature was varied to compensate and maintain the
same IMEP. The result was an engine condition reached through differing contributions
from of NVO and intake heating. An optimization point could then be found where total
PM mass was minimized. Because this study was conducted on a DI-HCCI engine
34
many of the differences in PM signatures are attributed to subtle changes inevaporation
and fuel mixing effects. Most interesting is the documentation and explanation of the
simultaneous reductions in total PM mass and NOX for a fixed speed and IMEP. As
EGR rates were increased through delayed IVO, in essence limiting intake air for a
fixed amount of residual thus increasing overall EGR percentage, notable drops in both
PM and NOX could be seen. The reductions in PM are explained by the fuel
experiencing more mixing time in the hot recompressed residuals, prior to the
introduction of intake air. Insight into NOX reductions is gained through the maximum
rate of in-cylinder pressure rise for these cases. Lower pressure rise rates are
characteristic of longer duration and lower maximum temperature combustion which
limits NOX production rates.
In summary, the authors of this work showed decreasing particulate emissions with
increasing intake air temperature. The effect was attributed to enhanced fuel
evaporation due to higher in-cylinder temperatures during compression. Because
evaporation and wall wetting phenomena are unique to DI fueling strategies, it should
be noted that this trend is not expected to be characteristic of all HCCI engines.
Additionally trends of NOX and PM emissions both decreasing with increasing EGR
were reported. These trends differ drastically from the widely accepted PM – NOX
tradeoff found in the Diesel emissions literature, but show good agreement with the
simultaneous PM and NOX reductions associated with HCCI combustion. Finally, an
overall trend of increasing loads and speeds leading to increased PM emissions, as is
also characteristic of both CI and SI engines, was shown.
A second publication by Misztal et al. (2009b) examines the role of injection
timing in PM formation in the same engine as described above. This work again
examines HCCI emissions when operating at very rich fuel to air ratios in terms of
HCCI operation, on the order of λ = 1. The emissions data collected were compared to
data obtained from the same engine operating in DISI mode. HCCI data is presented
from both a single injection strategy and a split injection strategy. In terms of general
HCCI operating conditions λ = 1 is very rich and represents very high loads. As with
35
previous work by Misztal et al., this results in significant accumulation mode particulate
matter (soot).
The DI mode of fuel delivery led to a high sensitivity of PM formation to injection
timing. Injection timings were reported in terms of end of injection (EOI) and were
varied from 250º BTDC to 350º BTDC of the compression stroke. The most advanced
timings generally showed the highest PM mass and number emissions even though
mixing times were the longest. The authors attributed this to wall wetting effects from
impingement of most of the fuel on the piston surface. As timings were delayed,
consistent reductions in PM were reported until the trend eventually reversed, this was
explained by a lack of mixing time prior to ignition. Optimum timings were found to be
those in unison with the intake valve opening event. This relationship was explained by
the increased mixing from the induction of fresh air into the cylinder. The authors
conclude that PM emissions are very closely coupled to mixture homogeneity for this
type of HCCI engine.
Although difficult to achieve, the purest form of HCCI requires fully premixed fuel
and air. All of the above PM emissions studies were done with DI fueling systems and
as a result show effects of diffusion burning. In order to clearly understand PM
formation in HCCI and other low sooting engines it will be very beneficial to examine
the most basic cases of premixed compression ignition combustion and employ a fully
premixed charge.
36
Chapter 3
PM Emissions Instrumentation
The sensitivity of particulate emissions measurements to sampling conditions and
practices has been examined by numerous authors. Recently Kasper (2005) and Mohr
(2005) have evaluated PM emissions sampling practices and examined the magnitudes
and sources of possible error. As engine manufacturers are forced to build cleaner
engines, the absolute magnitude of PM emissions, both number and mass, falls rapidly
leaving the quantities of interest closely approaching noise levels. Kasper (2005) notes
in particular the advantages found in both sensitivity and time resolution with modern
nanoparticle sampling instrumentation compared to gravimetric analysis.
With HCCI combustion, emissions of PM, or smoke, are frequently described as
“near zero” (Juttu, 2007) or “ultra low” (Epping, 2002). However recent work by Kaiser
et al. (2002), Price et al. (2007), and Misztal et al. (2009a and 2009b) have shown that
although the total mass of PM is indeed drastically reduced, significant numbers of
particles remain in the size ranges below 100 nm in mobility diameter. This size range
is well within the capabilities of modern nano-particle sampling instrumentation, which
is well suited for studies in HCCI PM emissions.
3.1 Size Distribution Characterization
Characterization of combustion exhaust particle size distributions has been
thoroughly research by a multitude of SI and CI engine studies. This work will draw
from the methodology of previous researchers, applying lessons learned to a new
problem, the emissions characteristics of HCCI engines. The primary suite of
instrumentation used consists of a Scanning Mobility Particle Sizer (SMPS),
Condensation Particle Counter (CPC), Engine Exhaust Particle Sizer (EEPS), and
Tandem Differential Mobility Analyzer (TDMA).
3.1.1
CPC
The most basic of these instruments is the continuous flow condensation particle
counter (CPC) which possesses only single particle counting capabilities. The basic
function of these instruments is to grow fluid droplets around particles initially too
37
small to detect by light scattering. The droplets are then counted with light scattering
methods. To accomplish the droplet growth, the aerosol is first passed though a region
of saturated vapor. It is then subjected to a temperature gradient in order to achieve
supersaturation of the vapor. Droplet growth initiated due to supersaturation, with the
small particles used as condensation nuclei for liquid droplets.
Two types of continuous flow CPCs exist, with the primary difference between
them the working fluid (water or butanol). Significant design differences are also
present. Continuous-flow, water based condensation particle counters (CPCs) were
introduced in 2003 (Hering, et al., 2005). Along with the obvious difference in working
fluid of the water-based CPCs, the thermodynamic approach for achieving
supersaturated conditions within the growth section also differs from that of a
conventional continuous-flow, butanol-based CPC. The ability of any CPC to activate
and grow a droplet around a condensation nuclei is characterized by the Kelvin equation
[13] (Hinds 1999). A given saturation ratio (SR) corresponds to a distinct particle size or
Kelvin diameter (d*) capable of maintaining mass equilibrium under the stated
conditions. Where Pd is the partial pressure at the droplet-vapor interface and PS is the
saturation pressure of the vapor. This quantity is defined by the material properties,
density (ρ), surface tension (γ), and molecular weight along with the temperature of the
system (T) and the universal gas constant (R). Above this diameter, mass flux to the
droplet surface is greater than away from it and the droplet grows to a size detectable by
light scattering.
SR =
 4γMW 
Pd
= exp 
* 
Ps
 ρRTd 
13
Figure 8 and Figure 9 illustrate the design of butanol and water based CPC growth
sections. These schematics are simplified versions taken from the work of Agarwal and
Sem (1980) and Hering et al. (2005).
38
QOUT
QIN
Condenser
Sample Flow
Saturator
Figure 8: Butanol CPC particle growth section
Conventional laminar flow CPCs, referred to from this point forward, as butanol
CPCs develop supersaturation by heating liquid butanol in the saturating region to a
point significantly above the temperature of the condensing region. The aerosol passes
through the saturating region and then enters the condensing region where a
thermoelectric cooling device drops the temperature, thus lowering the saturation
pressure of the system and leaving the previously saturated butanol vapor in a
supersaturated condition. This is made possible by rapid rate of thermal diffusion
relative to that of mass diffusion of butanol in air. We know that the Kelvin diameter of
a particle decreases with saturation ratio. Increasing supersaturation, by increasing the
temperature difference between the saturator and condenser, moves down the minimum
particle size that can be activated and grown. The maximum temperature difference is
bounded on the by the onset of homogeneous nucleation.
39
QIN
Q
Wetted Growth Tube
Sample Flow
Figure 9: Water CPC particle growth section
Water based CPCs use a different method to achieve supersaturation. As shown in
Figure 9, the aerosol flow initially enters a conditioning region where the temperature is
brought to a temperature below that of the growth region and the relative humidity is
brought to 100% by way of diffusive mass transfer from a saturated wall wick (Hering
et al. 2005). The aerosol then enters a heated growth region, where the saturation
pressure of water consequently becomes much higher. According to Hering and
Stolzenburg (2005) a region of supersaturation results along the centerline of the growth
tube, owing to the relatively high rate of mass diffusion of water vapor in air compared
with that of thermal diffusion of air.
Worst case performance evaluations by Hering and Stolzenburg (2005), Hering et
al. (2005), and Liu et al. (2006) reported differences in instrument response to sample
aerosols of differing composition. In the same studies instrument response improved
when examining challenge aerosols with increased hygroscopicity, thus better suited to
growth by water condensation. These studies conclude that most aerosols in real-world
environments would be adequately hydrophilic to allow proper particle activation and
growth. Field studies near roadways by Biswas et al. (2005) and Hering et al. (2005)
confirm the aforementioned hypothesis. More recently Mordas et al. (2008) and
Hermann et al. (2007) found similar results of increased D50 with highly pure
hydrophobic aerosols along with better performance with the addition of minute
hydrophilic impurities. Ambient atmospheric data from mobile emissions sources has
40
been presented by Biswas et al. (2005) and Hering et al. (2005). The work conducted
here was executed with butanol based CPCs due to availability.
3.1.2
SMPS
Initially developed by Wang and Flagan (1990), the SMPS utilizes a Differential
Mobility Analyzer or DMA (Knutson and Whitby, 1975) coupled with a CPC. The
principle of operation of the DMA is explained below and illustrated in Figure 10.
The DMA portion of the SMPS is used to classify particles by electrical mobility.
An electric potential is put on the center rod while the outer tube of the DMA is held at
ground. When operated with laminar sheath and aerosol flows a scenario is created
where a particle’s trajectory is established from two velocity components. The first, a
result of the particles drag, follows the sheath flow in the axial direction, and the
second, a result of the particles electric mobility, induced by the applied potential, is in
the radial direction. This induced force is balanced with the drag force working against
the particle in the radial direction. To predict the velocity in the radial direction the
following relations describing these two forces are equated. CC is the Cunningham slip
correction factor, µ is the dynamic viscosity of the fluid, dP is the particle mobility
diameter, and V is the particles terminal velocity in the radial direction.
neE =
3πµVd P
CC
14
Solving for velocity V and simplifying with mechanical mobility, we obtain the
following expression for particle velocity, influenced by a given electric field, in a
specific fluid.
V = neEB
41
15
Where n is the number of charges on the particle, E is the strength of the electric field, e
is a constant for the elementary unit of charge, and B is the particles mechanical
mobility. Hinds (1999) defines mechanical mobility as the relative ease of producing
steady motion for an aerosol particle. This is defined quantitatively by dividing the
terminal velocity of a particle by the drag force acting on it.
When the above velocity vector is added to the velocity vector of the sheath flow a
unique trajectory is defined for a unique particle mobility diameter. By adjusting the
applied field particles of differing mobility diameters can be given the same trajectory.
This is the desired function of the DMA. An exit path exists so that only particles
following the prescribed trajectory will make their way into the exit geometry. All
others will impact the center rod and stick or be flushed out with the excess air flow.
Sheath Air
Aerosol Flow
Excess Air
Sample Flow
Figure 10: DMA flow schematic
The DMA operating voltage range is scanned through via an algorithm controlled
high voltage power supply, with the CPC reporting concentrations at a mobility
42
diameter corresponding to each voltage. A data reduction algorithm then interprets the
CPC counts and gives a distribution of concentrations versus mobility diameters. A
commercial version of this software, provided by the instruments manufacturer, was
used.
Because different CPCs have different performance characteristics, particularly
counting efficiency, the software must account for these. Counting efficiency relates the
specific instrument response at a given particle size to the true concentration of particles
as measured by an external calibration source. As the lower response limit of the CPC is
encountered, counting efficiencies fall from near 100 percent to near zero in the span of
a few nanometers. The current version of the Aerosol Instrument Manager (AIM)
software allows for user inputs of operating parameters so that all pertinent variables for
SMPS operation are defined. The AIM software compensates for reduced counting
efficiency by referencing a counting efficiency curve for each CPC. This curve is a list
of efficiencies and corresponding particle diameters denoted with an “eff” file extension
in the AIM program files. The software includes unique counting efficiency curves for
each of the manufacturers particle counters that the software supports. If the particle
counters are following their expected behavior, the effects of counting efficiency should
be fully remedied by this correction. The AIM software also has a correction for
diffusion losses within the components and associated plumbing of the SMPS.
Brownian diffusion, driven by a concentration gradient between the aerosol stream and
the walls where concentrations are effectively zero, leads to significant losses of
particles smaller 100 nm as shown by Reineking and Porstendörfer (1986). The
diffusion correction employed by the software compensates for these losses. Aside from
particle size, residence time within the SMPS flow path is the only input affecting
diffusion losses. Each CPC has a specific flow rate and combining this with the defined
geometry of the SMPS enables the software to correct for diffusion losses. In principle,
these corrections enable the AIM software, regardless of instrument configuration to
theoretically calculate homogenous results. It is important to note that at very small
particle sizes where diffusion losses are most pronounced and counting efficiency of the
instrument is very low, the reported concentrations are highly corrected and based on a
43
few sporadic particle counts. This leads to a very high sensitivity to false counts and
highly inferred data sets which should be closely examined.
Two DMAs were available for use in the SMPS, a long DMA (TSI model 3081)
and a nano DMA (TSI model 3085). The nano DMA is optimized for size classification
of particles in the 3 to XX nm size range. The long DMA has a range of 10 to 1000 nm.
An SMPS comprised of a nanoDMA was used for the majority of the work conducted
here. With SMPS sheath and aerosol flows set to 15 and 1.5 lpm respectively for all
work conducted, the SMPS was operated with a size range of 3 to 64 nm. Preliminary
measurements indicated the absence of particles above this range. These flow rates were
selected to give as small as possible minimum particle diameter while maintaining a
size range that encompassed all expected particle sizes.
Number concentrations can then be converted to mass through a density and
volume calculation at each particle size. For all mass calculations a particle density of
1.0 g/cm3 was used. Schnieder et al. (2005) has shown this to be a reasonable estimate
of density for PM originating from engine lubricating oil. This density was also used for
PM studies on a gasoline fueled HCCI engine by Misztal et al. (2009a).
3.1.3
EEPS
Most recently developed of the particle instruments used is the Engine Exhaust
Particle Sizer (EEPS), which gives comparatively fast response, on the order of 10 Hz
(Johnson et al., 2004). The technology was developed at the University of Tartu and
commercialized by TSI Inc. Unlike the SMPS, the EEPS is an electrometer based
instrument and does not physically count individual particles. Particles entering the
EEPS are initially given a unipolar charge then routed through an annular space with an
electric potential applied to a center rod and the outer cylinder held at ground. The outer
cylinder of the annulus is comprised of a set of isolated electrometers. Each
electrometer corresponds to range of known electrical mobilities and corresponding
particle sizes. The instrument sample flow and particle terminal velocity due to
electrical mobility allow calculation of a bin of particle sizes that will be detected by
44
each electrometer. Particle concentrations at a given size are then inferred from currents
in corresponding electrometers.
Although the fast time response of the EEPS makes it a valuable tool for PM
emissions research, the work conducted here concentrated on steady state engine
conditions. In addition the EEPS has poorer sensitivity and size resolution that the
SMPS system. Consequentially, the SMPS was used much more extensively in this
work.
3.1.4
TDMA
Experimental work carried by Liu, et al. (1978), McMurry, et al. (1983), and Rader
et al. (1986) began using two DMAs in series as a means of further understanding
chemical and thermo physical properties of aerosol particles. Rader and McMurry
(1986) adopted the term TDMA to describe apparatus which feature two or more DMAs
operated in series. Application of TDMA to study droplet evaporation and growth has
been rigorously characterized by Rader and McMurry (1986). Orsini et al. (1998) have
also given a detailed description of adaptation of TDMA for measuring volatile
fractions of particles. The TDMA technique published by Orsini (1998) was
successfully used by Sakurai (2003) to study Diesel nanoparticle composition. A similar
system has been used for this work and is shown in Figure 11. The thermal conditioner
was sized in accordance with Orsini (1998) and Sakurai (2003), with temperature
monitored continuously at the conditioner outlet. The temperatures used for the thermal
conditioner ranged from 40° to 110°C with data collected in 10° increments.
45
Fixed Voltage
DMA
Source
Po 210
Source
Scanning Voltage
DMA
Po 210
Source
Ultrafine CPC
Thermal Conditioner
Sample
Aerosol
Figure 11: TDMA Apparatus
3.2 Dilution
To simulate the process of hot exhaust gases mixing with ambient air and cooling a
micro-dilution system very similar top the one developed by Abdul-Khalek (1999) was
used. The system draws a small sample of exhaust from the engine outlet and dilutes it
in two stages with a variable residence time aging chamber between. The conditions
pertinent to aerosol nucleation and growth by condensation are dilution ratio, dilution
air temperature, residence time in the aging chamber, and relative humidity. The microdilution system allows for easy manipulation and tight control of these variables. A
schematic of the system is shown in Figure 12.
46
Dilution Air
Regulator
2
Temperature
2 Control
Regulator
1
Water Out
Stage 2
Ejector Pump
Dilution Tunnel
CO2
Temperature
1 Control
Stage 1
Ejector Pump
Water In
CO2
Figure 12: 2 Stage Micro-Dilution System
CO2
A key point to be made here is that HCCI engine exhaust is thought to produce
significant number of particles with diameter smaller than 50 nm (Price et al., 2007;
Kaiser et al., 2002). When examining detailed characteristics of these particle size
distributions, great care must be taken to tightly control dilution conditions due to the
sensitivity of nucleation to dilution conditions as shown by Abdul-Khalek (1999). The
sensitivity of nucleation mode particles, or nanoparticles, to dilution system residence
time was clearly shown at nearly all dilution ratios and temperatures examined. As
residence time increased from 100 ms to 1000 ms, the number of nucleation mode
particles increase by more than an order of magnitude. Meanwhile accumulation mode
particle concentrations remained unchanged. A relationship between primary dilution
air temperature and nucleation mode behavior was also established and a trend of
falling concentration with rising temperature established. The sensitivity of
nanoparticles to dilution conditions was further explored by Abdul-Khalek et al. (2000)
and Mathis et al. (2004). Abdul-Khalek et al. isolated particle growth rates and reported
them to vary significantly with primary dilution temperature. Additionally strong trends
of decreasing growth rate with increased primary dilution ratio were established by
Abdul-Khalek et al. (2000). Additionally these authors showed strong trends of
decreasing growth rate with increased primary dilution ratio. Total concentration of
nucleation mode particles was again shown to steadily drop as primary dilution
temperatures were increased. Mathis et al. confirmed these results and reported a
47
maximum change of more than an order of magnitude exhibited when primary dilution
temperatures were increased from 17°C to 40°C. The effects of humidity were also
investigated. Mathis et al. found nucleation mode concentrations increasing by an order
of magnitude as humidity increased from 2% to 51% RH. Similar sensitivity to
humidity was shown via modeling of nucleation mode PM from Diesel combustion by
Kim et al. (2002). These authors noted number concentrations increasing by a factor of
6 when relative humidity of the dilution air was increased from 10% to 90%.
Both primary and secondary dilution ratios are set through air flow rates. For this
work they were also confirmed initially with NO concentrations at the selected dilution
conditions. Due to the sensitivity of nanoparticle formation to primary dilution
conditions, the primary dilution ratio was continuously monitored via CO2
concentrations (ppm) through Equation 16.
DRPrimary =
(CO
(CO
2, Exhaust
2, Primary
− CO2, Ambient )
− CO2, Dilution Air )
16
Figure 13 shows the sensitivity of ethanol HCCI nucleation mode PM to stage one
dilution air temperature. The data was collected with a stage one dilution ratio of 17.7:1,
a dilution tunnel wall temperature of 25°C, a stage two dilution ratio of 15:1, and a
stage dilution air temperature of 25°C. The engine was operated on pure ethanol fuel, at
a low load condition, with a speed of 1500 RPM. At low stage one air temperatures,
25°C, nanoparticle formation increased significantly. However, with stage one dilution
air temperatures between 30° and 40°C, formation of the nucleation mode remained
relatively stable. Variations in stage 2 dilution air temperature show little effect on
nucleation mode formation and are not presented here.
48
1.4E+09
S1 = 25°C
S1 = 30°C
S1 = 35°C
S1 = 40°C
S1 = 45°C
1.0E+09
3
dN/dlogdP (#/cm )
1.2E+09
8.0E+08
6.0E+08
4.0E+08
2.0E+08
0.0E+00
1
10
DP (nm)
100
Figure 13: PM variation with stage one dilution air temperature
In an effort to more precisely control dilution conditions within the aging chamber,
or tunnel, the outer wall was water jacketed and held at a constant temperature equal to
that of the stage one dilution air. The following analysis was conducted to ensure air
temperatures within the tunnel were reasonably close to tunnel wall temperatures.
Taken from Kaminski and Jensen (2005), Equation 17 predicts the exit fluid
temperature for a case of internal flow with constant wall temperature.
 hA
TExit = (TInlet − TWall )exp −
 m& c P

 + TWall

17
Where TExit is the mean fluid exit temperature, TInlet is the temperature of the fluid
entering the tunnel, TWall is the dilution tunnel wall temperature, h is the forced
convection coefficient for the system, A is the cross sectional area of the tunnel, m is
the mass flow rate of fluid through the tunnel, and cP is the constant pressure specific
heat of the fluid in the tunnel. A Reynolds number for the tunnel entrance was
49
calculated to be roughly 10,500, indicative of a turbulent flow condition. Combined
with a Prandtl Number of .696 for air at these conditions, a Nusselt number of roughly
30 was calculated which then allowed a convective heat transfer coefficient (h) equal to
5 W/m2 °C to be found. Figure 14 shows exit air temperature profiles at different
distances along the length of the 120 cm dilution tunnel, found through Equation 17.
Fluid Exit Temperature (°C)
60
50
Tw=45°C
Tw=40°C
Tw=35°C
Tw=30°C
Tw=25°C
Tw=20°C
40
30
20
10
0
0
25
50
75
100
125
150
Distance from Tunnel Inlet (cm)
Figure 14: Mean exit temperature profiles along the length of the dilution tunnel,
varied wall temperature, TIn = 50°C, Air flowrate = 80 slpm
An inlet temperature of 50°C was used and is based on a 15:1 dilution ratio of
300°C exhaust air with 35°C dilution air. It can be seen that at roughly half the tunnel
length (60 cm) the air temperature is within 3°C of the wall and at the end of the tunnel
(120 cm) the air temperature is only 1°C higher than the wall. Measured air
temperatures at the exit of the tunnel were within 1°C of predicted and within 2°C of
the tunnel wall.
A sensitivity analysis also was conducted to explore the effect of dilution tunnel
wall temperatures on particle formation, and find a temperature at which the exhaust
aerosol was most stable. For this analysis stage one dilution ratio was held at 17.7:1 and
stage two at 15:1. Stage two dilution air temperatures were held at 25°C. At each of two
50
dilution tunnel temperatures, PM data was collected at three stage one dilution air
temperatures. These results of these experiments are shown in
1.4E+09
S1=25°C, T =25°C
S1=35°C, T =25°C
S1=45°C, T =25°C
S1=25°C, T =35°C
S1=35°C, T =35°C
S1=45°C, T =35°C
3
dN/dlogdP (#/cm )
1.2E+09
1.0E+09
8.0E+08
6.0E+08
4.0E+08
2.0E+08
0.0E+00
1
10
DP (nm)
100
Figure 15: Sensitivity of PM formation to dilution tunnel wall temperature
Based on the above sensitivity analyses, it was found that stage one dilution air
temperatures of 35°C gave the most stable nucleation mode. Additionally particle
formation showed reasonable stability at tunnel wall temperatures of 35°C. Stage one
dilution air and tunnel wall temperatures were both held at 35°C throughout all further
testing. Stage 2 dilution temperatures were held at 25°C.
51
Chapter 4
Preliminary Modeling
Identifying the operating conditions suitable for HCCI combustion is a critical step
in the development an experimental apparatus. CHEMKIN® software was utilized to
model the chemical and thermodynamic behavior of an idealized system. Using the
engine parameters of the actual test engine, a single cylinder of the test apparatus was
modeled. The cylinder is modeled as zero-dimensional homogeneous reactor. Fuel and
air are assumed to be well mixed and in the gas phase with both the intake and exhaust
valves closed at the beginning of the simulation. The charge is modeled in a single zone
with the entire contents of the cylinder viewed as a continuum. A very similar approach
has been used by Ng and Thomson (2007) and Martinez-Frias et al. (2007). For the
preliminary modeling work presented, the system boundaries (cylinder walls) were
viewed as adiabatic. Pertinent engine characteristics are shown in Table 1.
Table 1: Physical characteristics of test engine
Engine Model
Bore (mm)
Stroke (mm)
Crank Length (c, mm)
Connection Rod Length (l, mm)
Isuzu 4HK-1TC
115
125
62.5
198
Single Cyl. Displacement (cm3)
Clearance Volume (cm3)
l/c
Compression Ratio
1298
74
3.168
18.5:1
Combustion reaction mechanisms have been developed and published by
researchers specifically for the fuels used in this work. The mechanisms consider
intermediate species formed as fuels are broken down and oxidized into the products of
combustion. Additionally, tabulated thermodynamic properties for the cylinder contents
are required for evaluation of cylinder temperature and pressures based on species
concentration and cylinder geometry. Thermodynamic properties and the chemical
kinetic mechanism for ethanol combustion were taken from the work published by
52
Marinov (1999). The hydrogen combustion thermodynamic properties and chemical
kinetics used were those developed by Conaire et al. (2004). A suitable mechanism for
modeling the behavior both fuels was not available however. To examine the effects of
dual fueling strategies a mechanism was developed that utilized kinetic data from both
of the above reaction sets. Removing the reactions involving hydrogen from the ethanol
mechanism of Marinov and replacing them with the more extensive set of reactions
encompassed by the work of Conaire et al. gave a reaction set that could more
adequately model combustible mixtures of ethanol and hydrogen. The combined
mechanism was validated by comparing the results of hydrogen and ethanol combustion
modeled independently with their respective mechanisms to hydrogen and ethanol
modeled with the combined mechanism. All modeling was done with a time resolution
of one half of one CAD. Table 2 shows the important combustion properties of the fuels
used in the modeling work, which will also be used in the proposed experiments.
Table 2: Combustion properties of ethanol and hydrogen fuels
Property
Chemical Formula
Molecular Mass (kg/kmol)
Lower Heating Value (kJ/kg)
Density @ 380 K (kg/m3)
Stoichiometric Air to Fuel Ratio
53
Hydrogen
H2
2.016
121,000
.06465
34.06
Ethanol
C2H5OH
46.068
29,700
1.477
8.94
140
H2 Mech
EtOH Mech
Combined Mech
Motoring
120
Cylinder Pressure (atm)
100
80
60
40
20
0
-90
-60
-30
0
30
60
90
Crank Angle Degrees After Top Dead Center (°ATDC)
Figure 16: Cylinder pressure traces of simulated HCCI combustion of hydrogen
fuel with λ=2, intake temp. of 355 K, and engine speed of 1000 rpm
The HCCI combustion process of hydrogen has been modeled using three different
mechanisms for the same set of engine conditions. From Figure 16 it is clear that the
ethanol mechanism does not properly describe hydrogen combustion. Start of
combustion, defined here as the point where the pressure trace deviates from the
motoring trace, is shown to be delayed by nearly 5 CAD when modeling HCCI
combustion of hydrogen with the ethanol mechanism compared to the results obtained
with the hydrogen mechanism. Although the ethanol of mechanism of Marinov has a
series of 25 reactions dealing with the combustion of hydrogen, the hydrogen
mechanism of Conaire et al. has a more detailed set 42 of reactions. The combined
mechanism, shown in blue, matches the results of the hydrogen mechanism nearly
perfectly. To ensure that the replacement of the hydrogen reaction equations did not
affect the overall performance of the ethanol mechanism, a series of ethanol HCCI
simulations were executed comparing the ethanol mechanism with the combined
54
mechanism. The results are shown in Figure 17. Very reasonable agreement between
the two mechanisms is also seen with ethanol HCCI combustion. A slight deviation in
the pressure traces can be seen near the start of combustion however the deviation is
less than 1 CAD and for this work can be viewed as negligible.
140
EtOH Mech
Combined Mech
Motoring
120
Cylinder Pressure (atm)
100
80
60
40
20
0
-90
-60
-30
0
30
60
90
Crank Angle Degrees After Top Dead Center (°ATDC)
Figure 17: Cylinder pressure traces of simulated HCCI combustion of ethanol fuel
with λ=3, intake temp. of 400 K, and engine speed of 1000 rpm
The most basic purpose for modeling the system is to isolate conditions feasible for
combustion in a particular engine. HCCI combustion is bounded on the upper end of the
load range by uncontrollable rates of pressure rise resembling knock and on the lower
end by misfire. The conditions selected for modeling and experimental work were
selected based on their relationship with full rated load of the engine. The IMEP
corresponding to maximum rated load for this engine is roughly 14 atm (1400 kPa), if
we assume ηM to be on the order of 90%, an IMEP of 16 atm (1600 kPa) will result.
55
Values of λ were selected to cover low and mid load operating conditions. A pressure
trace obtained through the model allows IMEP to be calculated for each λ via Equation
4. Table 3 illustrates the modeled λ values used, the corresponding IMEP, and the
percentage of IMEP at maximum rated load for the test engine.
A series of range finding simulations were first executed to ensure that each λ was
within the limits combustibility for the fuels. After isolating the desired λ values, the
response of the model to intake temperatures was investigated. An IMEP sweep was
modeled using a fixed fuel flow corresponding to the initial lambda at 370 K. The
intake temperature was then changed in 10 K increments. Increasing intake temperature
changes the amount of air going into the cylinder through ideal gas behavior, thus the
actual lambda falls as temperature rises. In order to maximize efficiency, IMEP is
maximized at a fixed fueling rate. For each of the four initial λ values shown in Figure
18, a range of temperatures from 370 K to 410 K was swept though.
Table 3: λ,
λ IMEP, and rated power relationship for thermal test conditions
λInitial
5.0
4.0
3.0
2.0
Max
λMaxIMEP IMEP
(atm)
4.87
3.78
3.89
4.71
2.92
6.10
1.89
8.74
T_in for Max
IMEP
(K)
380
380
380
390
56
% IMEP at
Rated Load
24%
29%
38%
55%
140
λ=5 - 4.5
Cylinder Pressure (atm)
120
100
80
60
Motoring
T_in = 370
T_in = 380
T_in = 390
T_in = 400
T_in = 410
40
20
100
80
60
Motoring
T_in = 370
T_in = 380
T_in = 390
T_in = 400
T_in = 410
40
20
0
0
-90
-60
-30
0
30
60
90
Crank Angle Degrees After Top Dead Center (°ATDC)
-90
-60
-30
0
30
60
90
Crank Angle Degrees After Top Dead Center (°ATDC)
140
λ=3 - 2.7
120
100
80
60
Motoring
T_in = 370
T_in = 380
T_in = 390
T_in = 400
T_in = 410
40
20
λ=2 - 1.8
120
Cylinder Pressure (atm)
140
Cylinder Pressure (atm)
λ=4 - 3.6
120
Cylinder Pressure (atm)
140
100
80
60
Motoring
T_in = 370
T_in = 380
T_in = 390
T_in = 400
T_in = 410
40
20
0
0
-90
-60
-30
0
30
60
90
Crank Angle Degrees After Top Dead Center (°ATDC)
-90
-60
-30
0
30
60
90
Crank Angle Degrees After Top Dead Center (°ATDC)
Figure 18: Pressure vs. CAD at 5 intake temperatures for each of 4 lambda ranges,
EtOH fuel, 1000 RPM
From the pressure traces shown it is clear that increasing the inlet temperatures
advances the start of combustion. The results presented here also follow intuitive
thought. For a fixed geometry piston and cylinder device, increasing the initial
temperature of the system will result in higher temperatures being reached earlier in the
stroke. This in turn leads to advancement the start of combustion. As the mixture
composition gets richer (λ decreases), it can be seen that the pressure traces tend to fall
below the motoring trace. Additionally, at the lowest temperature case in the λInitial=2
plot, the pressure trace indicates that the mixture does not even ignite. These
phenomena are due to the high specific heat of ethanol, which is more than twice that of
air. Thus the effect is less noticeable in leaner mixtures which contain far less ethanol.
The effects of hydrogen on ethanol HCCI combustion were also simulated using
the same model as above. In the following simulations, hydrogen proportions were
calculated based on a percentage of power input through Equation 18. The mole
57
fractions of each fuel with respect to total fuel were then varied in order to correspond
to 0, 5, 10, 15 and 20 % hydrogen energy supplementing the main ethanol fuel supply.
 KJ 
 kg 
MWH 2 
Total Energy (KJ ) = nH 2 LHVH 2 

 kmol 
 kg 
 KJ 
 kg 
MWEtOH 
+ nEtOH LHVEtOH 

 kmol 
 kg 
140
140
Cylinder Pressure (atm)
100
80
60
Motoring
0% H2
5% H2
10% H2
15% H2
20% H2
40
λ=4 - 3.98
120
Cylinder Pressure (atm)
λ=5 - 4.99
120
100
20
80
60
Motoring
0% H2
5% H2
10% H2
15% H2
20% H2
40
20
0
0
-90
-60
-30
0
30
60
90
Crank Angle Degrees After Top Dead Center (°ATDC)
-90
-60
-30
0
30
60
90
Crank Angle Degrees After Top Dead Center (°ATDC)
140
120
λ=3 - 2.97
100
80
60
Motoring
0% H2
5% H2
10% H2
15% H2
20% H2
40
20
120
Cylinder Pressure (atm)
140
Cylinder Pressure (atm)
18
λ=2 - 1.96
100
80
60
Motoring
0% H2
5% H2
10% H2
15 % H2
20% H2
40
20
0
0
-90
-60
-30
0
30
60
90
Crank Angle Degrees After Top Dead Center (°ATDC)
-90
-60
-30
0
30
60
90
Crank Angle Degrees After Top Dead Center (°ATDC)
Figure 19: Pressure vs. CAD with varying hydrogen proportions for each of 4
lambda ranges, EtOH base fuel, 1000 RPM, intake temperature of 380 K
In these cases we also have a changing cylinder volume left for air with constant
fueling as the fuel proportions vary. The variance here is due to the discrepancy in the
energy to volume ratios between hydrogen and ethanol. As hydrogen energy is slightly
increased, the additional volume required for fuel displaces air and causes a decrease in
58
λ. From the simulations it can be seen that SOC advances with increasing hydrogen
proportion. It is also evident that IMEP is increased although the fuel energy input is
held constant. Table 4 shows the optimized IMEP cases for the variable hydrogen
energy input simulations. In the two leanest cases, the rate of pressure rise shows a
significant increase with increasing hydrogen energy. This behavior agrees well with
the findings of Yap et al. (2004), who examined similar amount of hydrogen addition to
natural gas HCCI combustion.
Table 4: λ, IMEP, and rated power relationship for hydrogen test conditions
λInitial
5.0
4.0
3.0
2.0
Max
λMaxIMEP IMEP
(atm)
4.99
3.72
3.98
4.59
2.98
5.98
1.96
8.59
% H2 for
Max IMEP
% of IMEP at
Rated Load
20
20
15
20
23%
29%
38%
54%
A third set of simulations were run in order to understand the effects of EGR on
ethanol HCCI combustion. Again a constant fueling rate was held for each lambda
range. Because intake air is being displaced by EGR in these experiments and the
fueling is held constant a new lambda is realized with each EGR rate. An intake
temperature of 380 K was used for each of the modeling runs. To obtain the proper
ratios of combustion reactants, a calculation was made to find intake charge mole
fractions from the contributions of fuel, air, and each of the EGR constituent molecules.
EGR rates (EGR) as a percentage of the total volume of intake air were used and the
amount of fueling was held constant. The only products of combustion to simulate EGR
were CO2, H2O, O2, and N2. These mole fractions were calculated at lambda values of
2, 3, 4, and 5 and EGR rates of 0, 10, 25, and 50 % using an EES code. The mole
fractions were then used as part of the input set for the CHEMKIN simulations.
A constant intake temperature of 380 K was used for all EGR simulations. In
practice this scenario is quite feasible. Although changes in intake charge temperature
59
will result from the introduction of hot exhaust gases, the intake heater feedback loop
samples temperature downstream of the EGR introduction port. As more thermal energy
is gained from the exhaust gases, the intake heaters can lessen their contribution and
only perform up to the level that is necessary to maintain the temperature set point.
140
140
λ =5 - 4.8
100
80
Motoring
EGR = 0%
EGR = 10%
EGR = 25%
EGR = 50%
60
40
120
Cylinder Pressure (atm)
Cylinder Pressure (atm)
120
100
20
80
Motoring
EGR = 0%
EGR = 10%
EGR = 25%
EGR = 50%
60
40
20
0
0
-90
-60
-30
0
30
60
90
Crank Angle Degrees After Top Dead Center (°ATDC)
-90
-60
-30
0
30
60
90
Crank Angle Degrees After Top Dead Center (°ATDC)
140
120
λ =3 - 2.8
100
80
Motoring
EGR = 0%
EGR = 10%
EGR = 25%
EGR = 50%
60
40
20
120
Cylinder Pressure (atm)
140
Cylinder Pressure (atm)
λ =4 - 3.8
λ =2 - 1.8
100
80
Motoring
EGR = 0%
EGR = 10%
EGR = 25%
EGR = 50%
60
40
20
0
0
-90
-60
-30
0
30
60
90
Crank Angle Degrees After Top Dead Center (°ATDC)
-90
-60
-30
0
30
60
90
Crank Angle Degrees After Top Dead Center (°ATDC)
Figure 20: Pressure vs. CAD with varying EGR rate for each of 4 lambda ranges,
EtOH fuel, 1000 RPM, intake temperature of 380 K
The simulations with EGR show peak pressures decreasing as EGR rates increase
for all lambda cases. The results agree well with the experimental efforts of Lü et al.
(2005b) for a high octane fuel. This is a clear result of the increase in mixture specific
heats as more CO2 and H2O are added. The above simulations show some dependence
of SOC on EGR rates in cases λ=2 and λ=3. However, in the λ=4 and λ=5 cases less
effect is seen on SOC and more of an effect on dP/dCAD materializes. This is
consistent with the findings of Au et al. (2001) in showing little effect on SOC and a
pronounced effect on dP/dCAD. The explanation behind this differing behavior with λ
60
range is unclear. The ability of EGR to limit peak rates of pressure rise is visually quite
clear in all of the cases except λ=2, where rates of pressure rise remain relatively
unchanged. Controlling rate of pressure rise is critical in maintaining smooth HCCI
combustion and ensuring knock is not encountered. Table 5 summarizes the peak
pressures obtained from the EGR modeling exercises.
Table 5: Summary of EtOH HCCI peak pressures with EGR, *indicates misfire
λInitial=2
%
EGR
0
10
25
50
Peak
Pres.
(atm)
113.5
107.4
93.6
47.3*
IMEP
(atm)
8.50
8.40
8.16
.23
λInitial =3
Peak
Pres.
(atm)
99.3
97.3
95.4
88.5
λInitial =4
IMEP
(atm)
5.97
5.83
5.90
5.81
Peak
Pres.
(atm)
88.3
87.0
84.9
80.2
IMEP
(atm)
4.56
4.56
4.53
4.49
λInitial =5
Peak
Pres.
(atm)
77.8
76.1
73.0
65.3
IMEP
(atm)
3.69
3.68
3.66
3.62
Summarizing the IMEP for the EGR conditions, it can be seen in Figure 20 that
increasing EGR rate both delays SOC and limits peak pressures. From these
characteristics of the pressure traces it is clear that increased EGR will lead to lower
IMEP. This is true provided the combustion trace does not deviate from the motoring
trace prior to TDC (0°) in the cycle.
An important assumption that must be justified for this modeling is that the mixture
is a thoroughly mixed combination of gases that can be viewed as a homogeneous
reactor. Theory of droplet evaporation and diffusion burning are important
considerations in Diesel engine (CI) combustion, however in the case of an HCCI
engine these issues are not of concern if the preceding assumption is valid. The
following analysis aids in the justification of this assumption. Derived from Fick’s law
of diffusion, Turns (2000) has developed Equation 19, which relates droplet evaporation
time(td) to initial droplet diameter(D0) via an evaporation constant (K).
61
D02
td =
K
19
Through Equation 20, K is defined by the following conditions of the system: droplet
density (ρDroplet), surrounding fluid density (ρ), diffusion coefficient (DAB), and the
transfer number, BY. The transfer number relates the mass fractions of vapor at the
droplet surface to those of the surrounding fluid and is given by Equation 21.
K=
8ρD AB
BY =
ρ Droplet
ln(1 + BY )
YEtOH ,S − YEtOH ,∞
1− YEtOH ,S
20
21
Through this analysis, the lifetimes of ethanol droplets were calculated in the intake
runner prior to induction into the cylinder via the valves. If, for simplicity we neglect
the latent heat of vaporization of ethanol, and an intake air temperature of 370 K and a
droplet temperature 10 K below the boiling point of ethanol are assumed, td in the intake
will range from .0001 to .04 seconds for 5 to 100 µm ethanol droplets. Residence times
(τ) of the fuel-air mixture before entering the combustion chamber have been calculated
based on intake geometry and engine speed and assuming the fuel is injected into the
moving air stream with the intake valve open. For engine speeds of 1000 RPM and
2500 RPM τIntake is equal to .018 and .007 seconds respectively. It is clear that
evaporation in the intake ports is not sufficient to ensure all of the fuel is in the gas
phase at the highest engine speed conditions. Further examining common practices in
port fuel injection mixture preparation, we see that the injector is minimally responsible
for atomizing fuel. Zhao et al. (1995) describes the design and operating characteristics
of modern PFI systems and states that fuel injections are usually timed while the intake
valve is still closed. The practice takes advantage of impingement of the spray on
interior intake surfaces to promote secondary atomization, giving droplet sizes generally
62
below 50 µm. Examining the window of time between intake valve closing from one
cycle and intake valve opening for the subsequent we can find a residence time of the
stagnant air inside the intake port of roughly .024 seconds at 2500 RPM. Summing
these two residence times with the in-cylinder evaporation time, roughly another .005
seconds at 2500 RPM, we can see that time for complete evaporation of even very large
diameter droplets is sufficient. In addition, the increased temperatures associated with
compression further enhance droplet evaporation. Droplet lifetimes were evaluated
during the compression stroke at cylinder temperatures and pressures found via
modeling. It is shown in Figure 21 that lifetimes fall by nearly 50% as temperatures and
pressures approach the SOC condition. These data help to ensure the ethanol droplets
will be completely evaporated upon SOC.
0.01
Droplet Lifetime (sec)
0.008
0.006
0.004
0.002
0
0
30
60
90
120
150
180
CAD Before Top Dead Center (°BTDC)
Figure 21: Lifetime of 50 µm ethanol droplets during the compression stroke of
Isuzu 4HK1-TC test engine, intake temperature is 370 K,
droplet temperature assumed to be 342 K (Tboil, EtOH = 352 K)
63
Chapter 5
Experimental Apparatus
5.1 Multi-cylinder Test Engine
Utilizing a multi-cylinder engine greatly increases the level of difficulty in
maintaining uniform HCCI combustion over a single cylinder engine. Variations incylinder temperature and charge composition can be caused by: coolant paths, EGR
distribution, intake air paths, and charge distribution. In order to precisely control as
many of these variables as possible a custom port fuel injection (PFI) intake manifold
was designed to replace the stock manifold. The multi-cylinder test engine with
modifications is shown in Figure 22.
The engine is based on a production Isuzu 4HK1-TC Diesel engine. The 4HK1-TC
is a 4-cylinder, 5.2 liter, turbo charged, direct injection engine. It was originally
equipped with common rail fuel injection and met all current emissions standards at the
time of production. A series of modifications has been made in order to convert the
Diesel engine to an HCCI engine. First the common rail fuel injection system was
removed. The injector rail, high pressure pump, and fuel distribution piping are not
required and were removed to clear space for new intake manifold. The original intake
manifold was very simple and had no separation between cylinder ports. This is an
acceptable design for a Diesel engine where only intake air and EGR flows through the
manifold, however it is not suitable for HCCI operation. Additionally the stock EGR
delivery system was removed, as it was poorly designed for ensuring even EGR flow to
each cylinder. The stock 18.5:1 compression ratio, piston design, and valve timings
were all maintained.
64
Thermal Management
System
EGR Distribution
Intake Manifold
H2 Fuel Rail
EtOH Fuel Rail
Figure 22: Multi-cylinder test apparatus
5.1.1
Intake Manifold
The design of the intake manifold allows for fully independent and isolated control
of fuel delivery, charge temperature, and % EGR for each of the four cylinders. This
isolated and independent control gives the opportunity to normalize in-cylinder
inhomogeneities across the four cylinders in terms of cylinder pressure or ignition delay
with one intake variable, while examining the effects of manipulating the other
available variables. The intake manifold, shown in Figure 22, couples all of the required
65
modifications for this HCCI engine. Two sets of fuel injectors, liquid and gaseous, EGR
ports, and thermally conditioned air ports are all designed into the manifold.
5.1.1.1 Fuel Injection
The fuel injection system is based on a standard PFI system with four
independently controlled, fully variable ethanol injectors. In addition to four liquid fuel
injectors, the system is equipped with a supplementary set of four gaseous hydrogen
fuel injectors with the same independent control. In order to minimize any effects of
intake wall wetting and maintain precise control of λ, the ethanol injectors are oriented
so that their spray pattern is focused on the intake port of the hot engine block. They are
timed to inject in advance of intake valve opening, giving adequate time for droplet
evaporation. The ethanol, is burned with the assumption that it is completely evaporated
and fully mixed with air upon SOC. To help promote complete fuel evaporation, the
high pressure injector supply line is preheated to very near the 85º C ambient pressure
boiling point of ethanol prior to injection. Additionally, intake air temperatures are also
well above this boiling point. Calculation of droplet lifetimes at these conditions,
summed with droplet lifetimes during compression, indicated total residence times of
the droplet within the intake runner, cylinder head, and combustion chamber to be well
in excess of the required time for full evaporation of the fuel droplets. Ethanol fuel flow
rates are monitored continuously via mass change in the fuel supply reservoir.
The hydrogen fuel injectors also discharge directly into the respective cylinder
intake ports. These injectors are timed to inject during the intake stroke with the intake
valves open. Hydrogen fuel flow rates were monitored with a Sierra (Smart-Trak) mass
flow meter.
Fuel flow rate for both injector sets can be controlled with injector pulse width and
with fuel supply pressure. Fuel injection is timed off the crank and cam shafts of the
engine and controlled via an aftermarket engine management system (Intelligent
Controls, IC 5420).
The undenatured ethanol fuel composition is shown in Table 6. The hydrogen fuel
was UHP/Zero grade with a purity of 99.999%.
66
Table 6: Ethanol Fuel Composition
Fuel Characteristic
Proof
Density @ 20 °C (g/cm3)
Water Content
Methanol
Specification
Typical
200
.7900-.7932
0.2 %
< 0.1%
200
.7904
0.2%
< 0.001%
5.1.1.2 EGR
EGR can take two forms, the first, external EGR consists of routing exhaust gases
from the exhaust side of the engine back into the fresh air intake. This strategy is
commonly used in both SI and CI engines as an emission control device. The second
form, where the gas components are often referred to residuals, maintains some level of
exhaust gases in the cylinder from the previous cycle. All engines have some level of
residuals remaining in the cylinder; however 2-stroke engines carry many more from
cycle to cycle due to the lack of dedicated exhaust and induction strokes.
External EGR on the test engine is manually controlled for each cylinder through a
set of four gate valves fed from the main EGR distribution manifold. A single branch of
the four cylinder EGR loop is shown in Figure 23. Temperature of the EGR manifold is
monitored along with exhaust back pressure and temperature. Pressure in the EGR
distribution manifold can be adjusted via the exhaust back pressure valve. By adjusting
the total back pressure the system, flow through the EGR manifold can be increased.
EGR flow to individual cylinders is controlled via an EGR throttle valve on each branch
of the EGR loop. Intake temperatures are held constant throughout varied EGR rates by
controlling the temperature of the fresh intake air stream. The EGR manifold was also
wrapped in an improvised water jacket in order to cool the EGR stream ensuring intake
temperatures could be maintained at constant levels.
67
Exhaust
Sample
Valve
Ambient
Air
EGR
Throttle
Valve
CO2 Analyzer
EGR Manifold
Intake
Temp
Intake
Sample
Valve
Exhaust Back
Pressure Valve
Intake
Air
Engine
Exhaust
Figure 23: Detail of EGR Loop
Accounting for ambient CO2 levels during testing, Equation 22 defines an EGR
rate by relating CO2 levels in the engine exhaust and the mixed stream of intake air.
EGR Rate (% EGR) =
(% CO
(% CO
2, Intake
2, Exhaust
− % CO2, Ambient )
− % CO2, Ambient )
22
5.1.1.3 Thermal Management
The thermal management system consists of independently PID controlled air
heaters upstream of the fuel injectors on each intake runner. Closed-loop feedback is
given via type K thermocouples located at the intake mounting flange. Prior to the
independently controlled heaters, a main set of preheaters elevate the temperature in an
initial step to ease the electric current burden on the independent heaters. The bank of
preheaters is designed to raise the intake temperature a maximum of 35°C above the
inlet temperatures when the engine is operating at a maximum of 3000 RPM. The
individual heaters are designed to then take the preheated air to the operating condition.
They are capable of creating a maximum ∆T of 95°C with the engine operating at 3000
68
RPM. This gives maximum intake temperatures on the order of 155° assuming an
ambient temperature of 25°C.
A recent study by Peineke et al. (2006) examined the use of glowing wires for
production of nanoparticles. In this series of experiments the wires were heated due to
resistance while passing a current through them and as a result metallic material is
sublimated. The wire was immersed in a flow of nitrogen and oxygen with temperature
controlled downstream of the heated wire via a water jacketed chamber. Utilizing the
cooled nitrogen/oxygen flow to induce nucleation, the authors reported size
distributions of highly pure metallic particles in the range of 3 to 80 nanometers
depending on the nucleation conditions and wire material. They also compiled a list of
suitable materials and developed an empirical rule for selecting useable material based
on saturation pressure at a given materials melting temperature. Although the precise
details of this series of experiments are not of particular interest in relation to the
proposed work, the phenomenon of nucleation of small metallic particles from a heated
wire is of extreme consequence. These particles, if present in the exhaust, could provide
nucleation sites leading to formation of exhaust PM around the metallic seed particles.
The exhaust aerosol would not be representative of an aerosol generated solely from
engine exhaust. Lee et al. (2006) noted significant differences in the concentrations of
metal bearing particles when examining emissions of an engine running on fuel doped
with ferrocene ((C2H5)2Fe). As ferrocene concentrations were increased from 20 ppm to
60 ppm, an order of magnitude increase was seen in the nucleation mode of the particle
size distribution. Additionally, metal particles were found via single particle mass
spectrometry throughout all particle sizes some doping levels. The authors explain the
presence of metal containing particles over the entire size range is likely due to three
formation and growth mechanisms, self nucleation of metal vapor, condensation onto
preexisting carbon particles, and coagulation of small metal containing particles.
The heating elements used in the thermal management system are nickel-chromium
(80%/20%) resistive heaters. With the heating element surface dimly glowing, surface
temperatures are likely to be in the region of 700-1000 K. At these temperatures
investigation is warranted into the production of particles by the resistive heaters. A
69
series of experiments was designed to simulate the temperatures and air flow rates the
system would encounter during engine operation. A variable speed blower was placed
upstream of the heaters simulate air flow normally provided by the engine. The
apparatus is shown schematically in Figure 24. The heaters in apparatus are the same
heaters as those used on the engine.
∆P
PABS
Filter2
Filter1
LFE
Blower
Heater1
(1500 W)
Heater2
(3700 W)
Sampling
Probe
T1
T2
Figure 24: Schematic of intake heating bench test
Flow rates were selected corresponding to engine operation at 1000, 1500, and
2500 RPM and measured with a laminar flow meter. At each of these flow rates outlet
air temperatures of 70°, 110°, and 150°C were examined and compared to the system
operating with no heat addition. Temperatures were set using closed loop PID control
with feedback from type K thermocouples T1 and T2 respectively. The 70°C data point
was chosen in order to examine the system operating only on heater 1, and the 150°C
point chosen as the highest expected temperature. PM data collected with an SMPS is
presented below. The SMPS sampled from the stainless steel sampling probe and was
used in anticipation of a nucleated particle size distribution. Upon testing it became
clear that no such size distribution existed and total concentrations are presented. The
70
process air was filtered initially at the blower inlet and again prior to the LFE. The
filters used were general purpose automotive air filters and are not HEPA grade, thus
some ambient PM is still present in the air stream. However, comparing the no heat (0°)
size distribution with those generated at 70°, 110°, and 150°C allows direct examination
of the particle generation of the heaters.
Figure 25 clearly shows no significant particle generation at any temperature or
flow rate. Total concentrations shown are on the order of 100 to 1000 particles per cm3.
For reference the total concentration of particles from 2.5-80 nm in the ambient lab air
at the time of testing was roughly 50,000 particles/cm3. In agreement with accepted
filtration theory, an increasing trend in total concentration can be observed as airflow
rate rises. These particle sizes are primarily captured by Brownian diffusion which is
less effective at high flowrates through the filter media (Hinds, 1999).
1500
1200
Concentration 900
(#/cm3)
600
300
0
H1:70°C, H2:150°C
1625 (2500 RPM)
H1:70°C, H2:110°C
975 (1500 RPM)
H1:70°C, H2:0°C
Heater Set Point (°C)
Air Flow Rate (lpm)
650 (1000 RPM)
H1:0°C, H2:0°C
Figure 25: Total concentration of particles between 2.5 and 80 nm at heater exit
71
Although the heaters were expected to give some PM addition, none was observed.
This is explained by the fact that the temperatures seen at the wire surface are far too
low to sublimate material. The melting point of the nickel-chromium alloy is
approximately 1400°C and the estimated temperature of the wire in the setup is much
lower, around 650°C. The work presented by Peineke et al. examined heated wires at
temperatures near the melting temperature of the material and noted that evaporation
rates are exponentially dependent on temperature. Particle concentrations reported by
those authors were on the order of 105 particles/cm3.
72
Chapter 6
Effects of Intake Temperature on
Emissions From an Ethanol Fueled HCCI Engine
The key to HCCI combustion is controlling the SOC without a physical ignition
event. Presented here is the first of a series of experiments designed to characterize the
emissions consequences of three SOC control strategies and their relationship to the
optimization of engine output, characterized by IMEP and BP. Specifically this work
explores thermal conditioning of intake air as a SOC control strategy and the emissions
effects realized due to altered combustion phasing.
Utilizing the procedures developed for characterizing nanoparticles from CI and SI
engines, data on particles from 3 to 64 nm in mobility diameter from an ethanol HCCI
engine will be presented. The instrumentation described above makes possible accurate
assessment of particle size distributions and allows meaningful conclusions to be drawn
for particles as small as 3 nm. A fully premixed fuel injection strategy will be used to
minimize the effects of droplet evaporation and diffusion burning. As noted by
Hyvönen et al. (2004), achieving congruent ignition conditions across multiple
cylinders in the same engine can prove very difficult. Variations in the gas exchange
process, compression ratio, cylinder cooling, fuel supply, and intake air temperature are
all present at some level. The test apparatus is well suited to the experiments, with three
easily manipulated variables available for control of SOC for individual cylinders:
intake temperature, % EGR, and fuel properties.
Adopted in the earliest research efforts (Najt and Foster, 1983), manipulation of
intake temperature to control SOC and extend the HCCI load and speed range has been
used extensively in many research activities since. The evolution of fast response
thermal management systems such as those reported by Flowers et al.(2005),
Haraldsson et al. (2004), and Peng et al. (2007) gives credence to the use of thermal
management as a means for controlling SOC under rapidly changing engine conditions.
In general, auto-ignition of low cetane fuels requires either unreasonably high
compression ratios or preheating of the intake air. Additionally, low load demand for
73
intake heating in HCCI applications stems from low in-cylinder temperatures which
limit the available thermal energy to the system.
The work presented examines the effects seen in both gas phase and PM emissions
over a range of loads. At each load condition, intake temperature was varied while
holding all other variables constant in order to maximize IMEP. At each intake
temperature sub-condition in this optimization process, emissions measurements were
made and then related to engine output. Intake temperatures were set and monitored via
PID controlled intake heaters. Small variations in intake temperature required for
congruent combustion across all cylinders were achieved through the independent
intake air heaters. A similar biasing process was developed and successfully executed
by Flowers et al. (2005) on a 6-cylinder HCCI engine. This was however, a proof of
concept exercise and no emissions work was conducted. Contributions of PM from the
heaters have been shown to be insignificant. There are two main goals for this series of
experiments. The first is to add to the understanding of PM emissions in HCCI engines.
Systematic research was carried out stepping though a set of predefined engine
operating parameters. At each condition the engine was allowed to reach steady state
and then emissions data were gathered. Resulting from this work is a map of PM
characteristics for a high octane biofuel, ethanol, when thermal conditioning is used to
control SOC in a HCCI engine. The second goal of this work will be to establish
operating conditions that will give stable operation for subsequent testing with fuel
blending and EGR used as combustion modifiers. Table 7 outlines the work executed
for this set of experiments. The conditions were chosen based on the manufacturers
speed corresponding to maximum rated torque with loads representing low and
moderate engine operation.
6.1 Experimental Procedure
A significantly modified 2005 5.2 liter Isuzu engine (model 4HK1-TC) is used for
these studies. The focus of the experiments is to explore the overall effects and
mechanisms in which altering SOC via thermal management is manifested in exhaust
emissions. The thermal management system developed for the engine, described in
74
detail in Section 5.1.1.3, provides a means to precisely control intake air temperature to
each of the four engine cylinders independently. The fuel is 200 proof, undenatured
ethanol. It is burned with the assumption that the fuel is completely evaporated and
fully mixed with air upon SOC. To help promote complete fuel evaporation, the high
pressure injector supply line is preheated to very near the 85º C ambient pressure
boiling point of ethanol prior to injection. Additionally, intake air temperatures are also
well above this boiling point. Calculation of droplet lifetimes at these conditions,
summed with droplet lifetimes during compression, indicated total residence times of
the droplet within the intake runner, cylinder head, and combustion chamber to be well
in excess of the required time for full evaporation of the fuel droplets.
The experiments were conducted with constant fueling, as a result, changing intake
air temperature led to small changes in the mass based λ. This is due to the volumetric
nature of air induction into the naturally aspirated engine, and as a consequence, λ
values are not constant but rather presented as a narrow range. A single engine speed of
1500 RPM was used. This speed corresponded to the manufacturers specified engine
speed at rated torque, it was selected to ensure smooth, well balanced engine operation
free of resonant vibrations. Table 7 summarizes the three load conditions explored, the
intake temperature ranges used, and the performance characteristics of each range.
Table 7: Thermal Management Test Conditions
λ range
Fueling Rate
(gEtOH/sec)
Fuel Input Energy
Rate (kW)
Intake Temperature
Range (°C)
Load Range (N•m)
IMEP Range (kPa)
Low
Engine Load
Mid-1
Mid-2
5.0 - 4.2
4.0 - 3.5
3.2 – 3.0
1.43
1.84
2.24
42.5
54.6
66.5
110-160
90-130
90-110
48-55
221- 236
59-93
233 - 318
118-128
383 - 403
75
At each of the loads, data were collected at 10°C intervals within the intake
temperature ranges listed in Table 7. In each case, further temperature increases were
halted as combustion advanced well before TDC and rates of pressure rise rose sharply.
A drop in output power coupled with obvious audible increases in engine noise
signified the onset of knock. The low end of the temperature range was bounded by
misfire, indicated by intermittent losses in engine output power.
In addition to collection of data on particulate emissions as described in Chapter
Chapter 3, gas phase emissions data was also collected. During all testing oxides of
nitrogen (NOX), carbon monoxide (CO), carbon dioxide (CO2), and unburned
hydrocarbons (HC) were monitored. Gas phase emissions data were collected with
conventional combustion gas analysis instruments. A California Analytical instruments
model 600-HCLD NOX analyzer was used for all NOX measurements. The instrument
measured wet emissions concentrations and was operated with a range of 0-10 ppm. HC
emissions data, also sampled wet, were collected with a J.U.M. Engineering 3-300A
hydrocarbon analyzer. CO and CO2 data were collected dry and corrected via a wet-dry
correction factor. The instruments used were a Horiba VIA-510 CO analyzer with an
operating range of 0-5000 ppm, and a Rosemont 880 CO2 analyzer with an operating
range of 0-15%. For continuous monitoring of dilution ratios, a second CO2 analyzer,
Sable Systems model CA-10, sampled exhaust gas downstream of stage one dilution
and was used to calculate dilution ratios.
Further range finding data was collected based on the conditions presented in Table
7. The engine exhibited stable operation from idle to loads up to 130 Nm. At a fixed
load of 90 Nm and a fixed intake temperature of 110 °C, the apparatus operated
smoothly at speeds ranging from 1000 RPM to 2250 RPM. Although engine operation
was stable at 2250 RPM, oil temperature slowly increased with engine speed and
surpasses 115°C at this condition.
6.2 Results and Discussion
The experimental results obtained while exploring the effects of intake air
temperature on the performance and emissions of an ethanol HCCI engine are presented
76
below. Initially an analysis of combustion phenomena is conducted, followed by an
emissions analysis. The results of both are compared with the literature and
relationships between combustion properties and emissions in a dual fuel HCCI engine
are then established.
6.2.1
Combustion Analysis
Looking first at the optimization of engine output, quantified here by IMEP, Figure
26 shows clear peaks for each of the load conditions where increased or decreased
intake temperature results in a loss of output power. IMEP calculations are derived from
the average IMEP across the 4 engine cylinders, each of which is based on 40 cycle
averaged in-cylinder pressure data. Error bars represent the standard error of the mean
for IMEP data across the 4 cylinders of the engine. BMEP is plotted on the same graph
with the position of maxima in agreement with the calculated IMEPs as expected. By
comparing an indicated parameter, derived from cylinder pressure, with a direct and
external measure of engine output, BMEP, we are able to qualitatively validate the incylinder pressure data acquisition and reduction methods used.
Low Load IMEP
Mid Load 1 IMEP
Mid Load 2 IMEP
Low Load BMEP
Mid Load 1 BMEP
Mid Load 2 BMEP
IMEP (kPa)
400
300
400
300
200
200
BMEP (kPa)
500
100
100
0
80
100
120
140
Intake Temperature (°C)
160
0
180
Figure 26: Optimization of engine output with intake temperature, ethanol HCCI,
constant fueling, 3 loads, 1500 RPM
77
Further investigation into the in-cylinder behavior will aid in understanding the
phenomena producing the changes in engine output. Figure 27, Figure 28, and Figure 29
show the average in-cylinder pressure behavior for the three load conditions as intake
temperature is varied. All pressure traces shown represent forty cycle individual
cylinder averages, again averaged across the four engine cylinders for a total of 160
cylinder cycles. Error intervals shown represent the standard error of the mean
calculated across the four engine cylinders. In agreement with the literature and
preliminary modeling, distinct advances in SOC are obvious as intake temperatures
increase. Peak pressures also increase with intake temperature at each of the three loads
tested.
Also shown in Figure 27, Figure 28, and Figure 29 on the right hand y-axis is heat
release rate. In calculating HRR, a single zone model similar to that presented by Stone
(1999) is used. The cylinder contents are assumed to behave as ideal gases composed of
an initially specified fuel and air mixture. The rates reported are net heat release rates
and neglect heat transfer to the cylinder walls. Heat release analysis allows quantitative
calculation of SOC timing, defined by the crank angle at which 10% of the heat energy
of the fuel has been liberated (CA10) and burn duration CA90-CA10. Additionally incylinder temperature is calculated from the heat release analysis.
Compared with modeled behavior, much slower pressure rise as combustion occurs
is shown here. Through these differences, the limitations of the model are made
obvious. The sharp rates of pressure rise exhibited by the model are due to its single
zone assumption with heat transfer neglected. In reality, wall cooling plays a significant
role in absorbing energy from combustion and all reactions are not taking place in a
perfectly simultaneous manner. Transfer of energy through the cylinder walls to the
engines coolant system leads to temperature distribution with cooler areas igniting later.
This leads to the slower rates of pressure rise seen in empirical in-cylinder pressure
data.
78
8000
220
T=110
T=120
T=130
T=140
T=150
T=160
Motoring
Pressure (kPa)
6000
5000
180
140
4000
100
3000
60
HRR (J/CAD)
7000
2000
20
1000
0
-20
-30
-20
-10
0
10
Crank Angle (°ATDC)
20
30
Figure 27: In-cylinder pressure behavior of ethanol HCCI combustion, fixed
fueling, λ =5.0-4.2, 1500 RPM, varying intake temperature
8000
260
T=90
T=100
T=110
T=120
T=130
Motoring
Pressure (kPa)
6000
5000
220
180
140
4000
100
3000
HRR (J/CAD)
7000
60
2000
20
1000
0
-20
-30
-20
-10
0
10
Crank Angle (°ATDC)
20
30
Figure 28: In-cylinder pressure behavior of ethanol HCCI combustion, fixed
fueling, λ =4.0-3.5, 1500 RPM, varying intake temperature
79
9000
300
T=90
T=100
T=110
Motoring
7000
260
220
Pressure (kPa)
6000
180
5000
140
4000
100
3000
HRR (J/CAD)
8000
60
2000
20
1000
0
-20
-30
-20
-10
0
10
Crank Angle (°ATDC)
20
30
Figure 29: In-cylinder pressure behavior of ethanol HCCI combustion, fixed
fueling, λ =3.2-3.0, 1500 RPM, varying intake temperature
Shown in Table 8 is a summary of the parameters used to characterize the
combustion process. These data were calculated from the same 160 averaged incylinder pressure cycles shown in Figure 27, Figure 28, and Figure 29. It can be clearly
seen that at all engine loads studied elevating intake temperatures leads to advances in
SOC. More advanced combustion in turn leads to higher peak heat release rates. With
more heat released prior to or very near TDC, the physical volume in which the energy
is released becomes smaller, and due to engine geometry, does not change as much per
CAD. This causes higher cylinder pressures and consequentially higher temperatures.
The behavior is clearly represented at each load. Variability of the combustion data was
shown to be low with the standard error of the mean in peak pressure timing calculated
across the four cylinders of the engine ranging from .10 to .18 CAD. The coefficient of
variation of the IMEP data across the 4 cylinders ranged from 0.5% to 5.3 %.
80
Table 8: Summary of combustion properties, ethanol HCCI with varying intake
temperature, 1500 RPM, 3 loads
Intake
Burn
Peak
Peak
IMEP
SOC
Temp.
Dur.
HRR
Temp.
(kPa) (ºATDC)
(°C)
(CAD) (J/CAD)
(K)
230
-2
10
80
1380
110
230
-3
8
90
1430
120
240
-5
7
100
1500
Low
130
Load
230
-6
6
100
1500
140
230
-8
5
100
1540
150
220
-9
6
110
1560
160
230
5
12
60
1220
90
310
1
9
110
1450
100
Mid
Load
320
-2
7
130
1546
110
1
310
-4
5
140
1590
120
300
-6
5
150
1630
130
380
5
10
130
1440
90
Mid
Load
400
0
6
170
1630
100
2
380
-3
5
180
1690
110
Two efficiencies characterizing engine performance are shown in Figure 30. The
first is combustion efficiency (ηCombust), calculated via Equation 23 from exhaust gas
components, inlet fuel flow, and intake air flow. It represents a measure of the unused
chemical energy carried out of the engine via the exhaust stream. In Equation 23, yi is
the exhaust gas mass fraction of each combustible species, hC,i is the heat of combustion
(LHV) for the given species, and m represents inlet mass flows of fuel and air.
Hydrogen, carbon monoxide, and unburned hydrocarbons are considered in this
analysis. Although PM contains combustible materials, it was neglected. PM mass
concentrations in the exhaust gas are roughly three orders of magnitude less than those
of CO and HCs resulting in a minimal contribution to combustion efficiency. The
second measure of efficiency shown for the engine is cycle efficiency, shown in
Equation 24, and defined as the indicated power output divided by the fuel chemical
energy input rate.
81
ηCombust = 1 −
∑y h
i C,i
23
 m& fuel


 hC,fuel
 m& air + m& fuel 
ηC =
IP
m& fuel hC,fuel
24
100%
50%
95%
45%
40%
90%
35%
30%
80%
Combust, Low
Combust, Mid 1
Combust, Mid 2
Cycle, Low
Cycle, Mid 1
Cycle, Mid 2
75%
70%
65%
60%
70
90
110
130
150
Intake Temperature (°C)
25%
ηCycle
ηCombust
85%
20%
15%
10%
5%
0%
170
Figure 30: Response of combustion and cycle efficiencies to variations in intake
temperature, ethanol HCCI combustion, 3 loads, 1500 RPM
As the engine intake temperatures are increased, the most notable jumps in
combustion efficiency are seen near the lowest intake temperatures. At these
temperatures the fuel conversion begins to deteriorate as in-cylinder temperatures, in the
coolest regions of the combustion chamber fall below those required for full oxidation
of the fuel.
82
6.2.2
Emissions Analysis
Figure 31 summarizes brake specific emissions of CO, HC, NOX, and PM for
ethanol HCCI at the low load operating condition. A total of 6 data sets were collected
at 10° C intervals ranging from 110°to 160°C. Number weighted particle size
distributions are shown in Figure 32 and mass weighted distributions in Figure 33. For
all mass calculations a particle density of 1.0 g/cm3 was used. Schnieder et al. (2005)
has shown this to be a reasonable estimate of density for PM originating from engine
lubricating oil. This density was also used for PM studies on a gasoline fueled HCCI
engine by Misztal et al. (2009a).
For all particulate matter distributions presented here, the error bars represent
confidence intervals established at the 90% level using a t-distribution, sample sizes
vary by data set from three to fifteen. Total levels of particulate number and mass show
significant sensitivity to intake temperature, with mass levels spanning nearly 3 orders
of magnitude. To clearly illustrate these variations and the details of the individual
distributions, particle mass data is shown on a log-log plot.
Following expected trends, brake specific CO and HCs at the low load condition
decrease as higher intake temperatures lead to higher in-cylinder temperatures through
advanced combustion. These thermal conditions, promoting more complete oxidation of
hydrocarbons, also lead to increasing NOX levels. Johansson (2007) describes the
thermal window above 1500 K, the temperature necessary to oxidize CO to CO2, and
below 1800 K the point at which NOX formation increases exponentially, as the optimal
operating arena for HCCI. Brake specific particulate emissions initially decrease with
intake temperature reaching a poorly defined minimum between 130 and 150°C, and
then increase. Referring to Table 8, the combustion parameters behave differently and
follow continued trends in the same direction throughout the temperature range at every
load. The reason for erratic PM emissions behavior at this condition is not fully
understood. However the particle mass distributions shown in Figure 33 are all similar
in shape with overlapping error bands. At the low load conditions distinguishing
between the magnitudes of the mass distributions with 90 % confidence is difficult due
to high variability.
83
BSCO
BSHC
BSNOx
BSPM
160
Brake Specific CO, HC
(g/kW hr)
140
120
0.3
0.25
0.2
100
0.15
80
60
0.1
40
Brake Specific PM, NOX
(g/kW hr)
180
0.05
20
0
100
110
120 130 140 150 160
Intake Temperature (°C)
170
0
180
Figure 31: Brake specific emissions from ethanol HCCI combustion with varying
intake temperature, fixed fueling, λ =5.0-4.2, 1500 RPM
2.5E+09
T=110
T=120
T=130
T=140
T=150
T=160
dN/dlogDP (part./cm3)
2.0E+09
1.5E+09
1.0E+09
5.0E+08
0.0E+00
1
10
DP (nm)
100
Figure 32: Mobility size distributions from ethanol HCCI combustion with varying
intake temperature, fixed fueling, λ =5.0-4.2, 1500 RPM
84
1.0E-01
T=110
T=120
T=130
T=140
T=150
T=160
dM/dlogDP (µg/cm3)
1.0E-02
1.0E-03
1.0E-04
1.0E-05
1.0E-06
1.0E-07
1
10
DP (nm)
100
Figure 33: Mass distributions from ethanol HCCI combustion with varying intake
temperature, fixed fueling, λ =5.0-4.2, 1500 RPM
The brake specific CO, HCs, and NOX emissions trends at the mid load 1 condition
are similar to those at the low load condition. Increasing intake temperature leads to
significant in-cylinder temperature increases through advanced combustion. This
promotes more complete hydrocarbon oxidation, but also promotes NOX formation. It
should be noted however that although NOX values show notable increases with intake
temperature, they remain extremely low over all conditions and intake temperatures
tested. The BSCO and BSHC values start out much higher than at the low load
condition, but as combustion temperatures increase, they fall dramatically until on par
with the low load levels. Lean burn HCCI combustion has excess oxygen available for
oxidation CO and HCs, but slow reaction rates prevent complete oxidation fuel due to
low in-cylinder temperatures.
Brake specific PM emissions are lower than at the low load condition, and like
NOX, increase strongly with intake temperature, opposite the trend shown for CO and
HCs. Particle number and mass distributions presented in Figure 35 and Figure 36
85
respectively show increasing concentration and size with increasing intake temperature.
Error bars show considerable fractional variability, especially at low temperatures.
0.2
BSCO
BSHC
BSNOx
BSPM
Brake Specific CO, HC
(g/kW hr)
160
140
0.16
120
0.12
100
80
0.08
60
40
Brake Specific PM, NOX
(g/kW hr)
180
0.04
20
0
80
90
100 110 120 130
Intake Temperature (°C)
140
0
150
Figure 34: Brake specific emissions from ethanol HCCI combustion with varying
intake temperature, fixed fueling, λ=4.0-3.5, 1500 RPM
86
2.5E+09
T=90
T=100
T=110
T=120
T=130
dN/dlogDP (part./cm3)
2.0E+09
1.5E+09
1.0E+09
5.0E+08
0.0E+00
1
10
DP (nm)
100
Figure 35: Mobility size distributions from ethanol HCCI combustion with varying
intake temperature, fixed fueling, λ =4.0-3.5, 1500 RPM
1.0E-01
T=90
T=100
T=110
T=120
T=130
dM/dlogDP (µg/cm3)
1.0E-02
1.0E-03
1.0E-04
1.0E-05
1.0E-06
1.0E-07
1
10
DP (nm)
100
Figure 36: Mass distributions from ethanol HCCI combustion with varying intake
temperature, fixed fueling, λ =4.0-3.5, 1500 RPM
87
The mid load 2 brake specific emissions data in Figure 37, show similar behavior
in brake specific CO, HC and NOX emissions to the previous two loads. Like the mid
load 1 number size distribution data, Figure 38 shows increases in number
concentration as intake temperatures are increased. However a decrease in mobility
diameter at an intake temperature of 110°C causes mass concentrations to decrease.
BSCO
BSHC
BSNOx
BSPM
Brake Specific CO, HC
(g/kW hr)
50
40
0.35
0.3
0.25
0.2
30
0.15
20
0.1
10
Brake Specific PM, NOX
(g/kW hr)
60
0.05
0
0
80
90
100
110
Intake Temperature (°C)
120
Figure 37: Brake specific emissions from ethanol HCCI combustion with varying
intake temperature, fixed fueling, λ=3.2-3.0, 1500 RPM
88
6.0E+09
T=90
T=100
T=110
dN/dlogDP (part./cm3)
5.0E+09
4.0E+09
3.0E+09
2.0E+09
1.0E+09
0.0E+00
1
10
DP (nm)
100
Figure 38: Mobility size distributions from ethanol HCCI combustion with varying
intake temperature, fixed fueling, λ =3.2-3.0, 1500 RPM
1.0E-01
T=90
T=100
T=110
dM/dlogDP (µg/cm3)
1.0E-02
1.0E-03
1.0E-04
1.0E-05
1.0E-06
1.0E-07
1
10
DP (nm)
100
Figure 39: Mass distributions from ethanol HCCI combustion with varying intake
temperature, fixed fueling, λ =3.2-3.0, 1500 RPM
89
The particle characterization work presented in Chapter 9 demonstrates that nearly
all the particles measured in this work were volatile. Furthermore, no measureable
concentration of solid accumulation mode particles has been observed. In the absence of
solid nucleation mode particles to act as adsorption sites the only gas to particle
conversion processes available for organic vapors in the exhaust as it dilutes and cools
are homogeneous nucleation and condensation. This leads the formation of volatile
nucleation mode particles. Tobias et al. (2001) reports that the composition of volatile
nucleation mode particles emitted by a CI engine shifts towards characteristics more
indicative of lubricating oil as loads increase. This suggests that higher in-cylinder
temperatures led to more lubricating oil related nucleation mode particle. For all loads
examined in the current HCCI study, the highest in-cylinder temperatures led to the
highest total mass of PM. Sakurai et al. (2003) later found that in diesel engines running
at light to moderate loads, volatile particulate matter was composed of at least 95%
compounds originating from unburned lubricating oil. Although different from standard
CI engines, HCCI engine have many characteristics in common; first, the charge is
ignited via compression, second, combustion generally occurs in a globally lean
environment, third, exhaust temperatures are usually much cooler than those found in SI
engine operation, and fourth, a similar configuration of the piston, piston rings, and
lubrication system is used. Thus delivery of lubricating oil to the combustion chamber
and its processing by the combustion system may be similar. With undenatured ethanol
used as the fuel in these tests, fuel contributions of sulfur, heavy hydrocarbons, trace
metals, and other impurities are virtually nonexistent. Although the HC emissions
relatively high, FTIR measurements showed them to be mainly unburned ethanol along
with smaller quantities of methane, ethylene, and low molecular weight aldehydes, all
of which are too volatile to condense under sampling conditions used here.
Furthermore, unburned HC decrease at higher intake air temperatures while PM
emissions increase.
Examining a hydrogen SI engine, Miller et al (2007) noted that as in-cylinder
temperature increased, organic carbon levels in the PM also increased. The authors
90
thought this was likely due to more complete breakdown of the oil lining the cylinder
walls. Data collected in this HCCI work suggests similar behavior is taking place here.
A multiple regression analysis was performed on all of the variable intake
temperature ethanol HCCI data to examine the dependence of total particulate mass on
select combustion parameters. The influence of the independent variables; SOC,
combustion duration, peak temperature, and peak heat release rate on the dependant
variable, total particulate mass, was analyzed. The size of the sample set used for
analysis was 14, which led an R2 value of .44. This relatively low R2 value reflects a
high degree of variability in the data. Examining the brake specific data presented in
Figure 31, Figure 34, and Figure 37 gives insight into the primary contributor to a high
degree of variability. The low load BSPM results, show in Figure 31 clearly exhibit a
high degree of variability with little apparent trend over the intake temperature spanned.
Comparing the R2 value for the entire data set with the one calculated without the low
load data included we see a significant change. The new data set had a sample size of 8
and yielded a R2 value of .95. These results indicate the possibility of errors in the low
load data set, the cause of which is unknown at this time.
6.3 Conclusions
Both gas phase and particulate brake specific emissions exhibit very clear
dependencies on intake air temperature. For each of the three loads tested, emissions of
CO and HC were highest at the lowest intake temperatures. These temperatures also
corresponded to points with significant losses in engine output power and efficiency,
suggesting incomplete combustion.
A dependence of total particle mass and number on intake temperature was
observed at all loads. Pure HCCI combustion is not base on flame propagation and
produces no locally rich burning, so that soot formation is generally avoided. With no
soot agglomerates acting as organic vapor sinks, these vapors act as precursors to
nucleation and lead to significant numbers of nucleation mode particles. Although some
variation is seen, in most cases the total particulate mass increases with peak HRR or
peak in-cylinder temperature. The dependence of total PM mass on peak temperatures
91
suggests that PM formation in fully premixed HCCI engines is associated with organic
carbon vapor from vaporized or atomized lubricating from the cylinder walls and piston
ring pack. Advanced SOC leads to higher heat release rates and higher in-cylinder
temperatures leading to elevated cylinder liner temperatures which increase the vapor
pressures and evaporation rates of organic compounds in the lubricating oil.
These results provide useful information on the influence of intake temperature at
various loads. However in later experiments it was found that particulate emissions
were lower at the same operating conditions as investigated here. The new findings
were found to be repeatable over all of the three loads. The subsequent testing spanned
three months and included multiple data sets. A possible explanation for these
differences is given below. The engine used in these test was originally configured and
used extensively as a conventional Diesel engine. The variable intake temperature
experiments were the first one conducted after the engine was converted to HCCI
operation. It is possible that the engine itself was not adequately broken in to remove
valve, ring, and piston deposits associated with its previous history. After a rigorous
period of operation at moderate loads and speeds up to 2250 RPM, the engine exhibited
the drop in PM emissions. After this drop the engine showed excellent repeatability
throughout all further experiments. All further data presented was collected after this
rigorous break in period.
92
Chapter 7
The Effect of EGR on Emissions in an
Ethanol Fueled HCCI Engine
Because EGR serves two purposes in HCCI engines, adding thermal energy to the
uncompressed mixture and acting as an energy sink to slow oxidation kinetics, it is
widely used for extending the HCCI operating range. Since the earliest studies by
Onishi et al. (1979), Noguchi et al. (1979), Najt and Foster (1983), and Thring (1989),
at least some level of EGR has been utilized in nearly all HCCI experiments. Recently,
work by Au et al. (2001) and Lü et al. (2005b) has contributed to the understanding of
SOC effects and burn duration effects of EGR. More modern applications like those
explored by Milovanovic (2004) take advantage of rapid response variable valve timing
to alter in-cylinder residual levels during transient engine operation. These applications
show promise for commercialization by altering effective compression ratio, EGR rate,
and ultimately SOC.
The goal of this work is to determine the influence of using EGR to control SOC
on the emissions from an ethanol fueled HCCI engine. PM emissions are closely
examined in order to explore the relationship between EGR and PM formation. Data are
presented on combustion behavior, gas phase emissions, and particulate phase
emissions and a relationship between EGR, combustion behavior, and emissions is
established.
7.1
Experimental Procedure
In conventional SI and CI engines, EGR has significant effects on emissions of
both PM and NOX. This work closely examines the relationship between EGR and PM
formation in HCCI engines. A global EGR rate is initially set at 0, 10, 25, and 50% of
intake air volume. This is accomplished by measuring CO2 concentrations in the
exhaust, the intake downstream of EGR mixing, and the ambient intake air. An EGR
rate is then calculated using Equation 22. The maximum output condition from the
thermal conditioning experiments was used as the starting point (0% EGR) for this
work. Consistent with the thermal conditioning experiments, fuel flows were kept
93
constant for each load throughout the EGR work. The feedback to the thermal
management system is located downstream of the EGR inlet allowing a constant intake
temperature to be maintained as the flow of hot exhaust gas to the intake was varied.
Although EGR cooling was in place, it is necessary to note that at the mid load 2
condition, with 25% and 50% EGR, the EGR cooler was not able to maintain
temperatures below the PID set point temperature. This resulted in elevated intake
temperatures at these conditions. Table 9 shows the test conditions examined. EGR
rates, realized lambdas calculated via carbon balance, and intake temperature set points
are shown.
Table 9: Test conditions for ethanol HCCI with varying EGR experiments
Low Load
Mid Load 1
Mid Load 2
TIntake = 130 °C
TIntake = 110 °C
TIntake = 100 °C
EGR Rate λC-Bal
EGR Rate
λC-Bal EGR Rate λC-Bal
0.0 %
4.35
0.0 %
3.63
0.0 %
3.06
1500
10.9 %
3.68
10.9 %
3.02
10.7 %
2.61
RPM
25.0 %
3.16
25.9 %
2.56
23.3 % *
2.10
50.9 %
1.83
49.3 %
1.53
48.8 % *
1.21
*Intake temperatures exceeded set point values due to EGR cooler limitations
The same modified 2005 5.2 liter Isuzu engine (model 4HK1-TC) as was used in
the thermal management work is employed for these studies. The engine was operated
at a speed of 1500 RPM. This speed corresponded to the manufacturers specified engine
speed at peak rated torque, it was selected to ensure smooth, well balanced engine
operation free of resonant vibrations. The same loads were investigated here as in the
previous thermal conditioning experiments. They were selected at reasonable intervals
representing low to moderate engine loads. The highest load condition was bounded by
the onset of engine knock. The fuel was 200 proof, undenatured ethanol.
These exercises required cylinder to cylinder biasing to establish a base level of
uniform EGR distribution and combustion across all cylinders. Fine tuning of EGR
rates was accomplished with independently controlled gate valves on the intake of each
94
cylinder. Precise measurement of individual EGR rates was possible via measurement
ports downstream of the EGR inlet on each intake runner, illustrated in Figure 23. Small
variations in individual cylinder intake temperature were dealt with through the subtle
use of the thermal management system. Because changing EGR rate affected CO2 levels
in the exhaust gas, an iterative approach was required to compute final EGR
proportions. Actual EGR levels were considered acceptable if they were within 10 % of
the target EGR proportion.
Due to lean operation of the engine, significant oxygen remains available in the
exhaust after combustion. In order to obtain actual lambda values, both fresh air and
oxygen originating from the exhaust must be accounted for. Müller (2010) presents a
method of calculating fuel to air ratio with the inclusion of oxygen from EGR. Internal
EGR, or residuals, are neglected for this analysis. The engine is a high compression four
stroke with modest valve overlap, leaving little residual gas in the cylinder from cycle
to cycle.
7.2 Results and Discussion
As perhaps the most universal tool for combustion control in HCCI engines, EGR
limits rates of pressure rise and controls peak pressures. Additionally, through the
addition of thermal energy to the fuel and air charge, it can also serve to advance
combustion. The experimental results obtained through varying the EGR rate to an
ethanol HCCI engine at three loads are presented here. Initially an analysis of
combustion phenomena is conducted, followed by an emissions analysis. The results of
both are compared with the literature and relationships between combustion properties
and emissions in a dual fuel HCCI engine are established.
7.2.1
Combustion Analysis
Figure 40, Figure 41, and Figure 42 show in-cylinder pressure behavior from the
three engine loads tested. The limiting effect of EGR on peak pressures and rate of
pressure rise is evident from examination of these figures. The increase in specific heat
of the mixture due to EGR is also made clear by the compression behavior documented
in these figures. In Figure 42, which is for the highest load condition, the limitations of
95
the EGR cooler are made clear through the details of the pressure behavior. The dual
thermal and chemical effects of EGR in HCCI engines are clearly evident. Initially, at
low EGR levels, the EGR cooler can maintain a constant intake temperature, allowing
documentation of purely the chemical effect of EGR. It then becomes apparent that
intake temperatures are increasing due to the SOC advance shown in the 25 and 50%
EGR cases. Measured intake temperatures for these cases were 110° and 140°
respectively for these conditions, compared to the 100°C set point used at this load.
Also shown in Figure 40, Figure 41, and Figure 42 on the right hand y-axis is heat
release rate. In calculating HRR, a single zone model similar to that presented by Stone
(1999) is used. The cylinder contents are assumed to behave as ideal gases composed of
an initially specified fuel and air mixture. The rates reported are net heat release rates
and neglect heat transfer to the cylinder walls. Heat release analysis allows quantitative
calculation of SOC timing, defined by the crank angle at which 10% of the heat energy
of the fuel has been liberated (CA10) and burn duration CA90-CA10. Additionally incylinder temperature is calculated from the heat release analysis.
220
8000
EGR=0%
180
EGR=10%
6000
EGR=25%
5000
EGR=50%
140
Motoring
4000
100
3000
60
HRR (J/CAD)
Pressure (kPa)
7000
2000
20
1000
0
-20
-30
-20
-10
0
10
Crank Angle (°ATDC)
20
30
Figure 40: In-cylinder pressure behavior of ethanol HCCI combustion with
varying EGR rate, fixed fueling, low load, 1500 RPM, 130° intake temperature
96
8000
260
7000
EGR=10%
6000
EGR=25%
220
180
EGR=50%
5000
Motoring
140
4000
100
3000
HRR (J/CAD)
Pressure (kPa)
EGR=0%
60
2000
20
1000
0
-20
-30
-20
-10
0
10
Crank Angle (°ATDC)
20
30
Figure 41: In-cylinder pressure behavior of ethanol HCCI combustion with
varying EGR rate, fixed fueling, mid load 1, 1500 RPM, 110° intake temperature
8000
260
7000
EGR=10%
6000
EGR=25%
220
180
EGR=50%
5000
Motoring
140
4000
100
3000
HRR (J/CAD)
Pressure (kPa)
EGR=0%
60
2000
20
1000
0
-20
-30
-20
-10
0
10
Crank Angle (°ATDC)
20
30
Figure 42: In-cylinder pressure behavior of ethanol HCCI combustion with
varying EGR rate, fixed fueling, mid load 2, 1500 RPM, 100° target intake
temperature
97
Table 10: Summary of combustion properties, ethanol HCCI with varying EGR
rate, 1500 RPM, 3 loads
Burn
Peak
Peak
EGR
IMEP
SOC
Dur.
HRR
Temp.
Rate
(kPa) (ºATDC)
(CAD) (J/CAD)
(K)
240
-5
7
100
1510
0%
240
-4
8
90
1540
Low
10%
Load
250
-4
8
90
1500
25%
240
-2
10
80
1600
50%
320
-1
6
120
1530
0%
Mid
320
0
8
120
1560
10%
Load
330
1
9
110
1520
25%
1
320
1
10
100
1650
50%
390
0
6
160
1590
0%
Mid
400
5
11
130
1540
10%
Load
410
4
9
140
1650
25%
2
390
-1
7
130
1820
50%
Table 10 summarizes the combustion properties calculated from in-cylinder
pressure data at each test condition. Variability of the combustion data was shown to be
low with the standard error of the mean in peak pressure timing calculated across the
four cylinders of the engine ranging from .18 to .34 CAD. The coefficient of variation
of the IMEP data across the 4 cylinders ranged from 1.0 % to 4.0 %. The role of EGR in
increasing burn duration and limiting peak rates of heat release is shown to be in good
agreement with the literature. Rahbari (2008) modeled the effects of EGR on ethanol
HCCI and found increasing EGR delays SOC, extends burn duration, and limits peak
cylinder temperatures. Reasonable agreement was also found with the work of Au et al.
(2001) and Lü et al. (2005b). However, these authors did not show significant delay in
SOC with increasing EGR rate. Dec et al. (2009) have found that in order to maintain a
constant MFB50, which refers to the point in the cycle where 50 percent of the fuel has
been burned, increasing intake temperatures were required as EGR rate was increased.
This translates into increasing EGR rates leading to delayed SOC at constant intake
temperature. Additionally Sjöberg et al. (2007) have clearly demonstrated a retarding
effect of EGR on ignition timing. The primary reasons for the effect are listed as; first,
the high specific heat of the gases reduces the compressed gas temperature, and second,
reductions of O2 concentration limits available O2 for combustion reactions.
98
Figure 43 shows the response of cycle and combustion efficiency to changes in
EGR at three engine loads. In all cases, cycle efficiency remains relatively constant
showing only a slight drop at the highest levels. A similar trend was recently
documented by Swami-Nathan et al. (2010) for varying EGR rates in an acetylene
fueled HCCI engine.
Combustion efficiency does show a slight decreasing trend as EGR rates are
increased at each load condition. EGR is employed in SI engines to act as a diluent and
soak up thermal energy during combustion (Abd-Alla, 2002). EGR should have a
similar effect in the cases studied here. For all loads shown, peak pressures fall as EGR
rates are increased. Because the combustion regime is near the low temperature limit for
the oxidation of CO to CO2, 1400-1500 K, lower in-cylinder temperatures lead to
increasing levels of CO in the exhaust, which is most prominently shown in the Mid
load 2 emissions data. This is one of the reasons for reduced combustion efficiency.
Emissions trends for CO and HCs at the Low Load and Mid Load 1 conditions do not
agree with the decreases in combustion efficiency shown in Figure 43. This is likely due
to the contributions of hydrogen in the exhaust stream to combustion efficiency.
Because hydrogen was not directly measured for the EGR experiments, the assumption
of a fixed water-gas equilibrium constant was used to calculate hydrogen emissions in
the carbon balance process outlined by Müller (2010). The increases in calculated
hydrogen emissions at higher EGR rates lead to decreases in combustion efficiency
which may be an artifact of this assumption.
99
100%
50%
45%
40%
95%
30%
90%
25%
Combust, Low
Combust, Mid 1
Combust, Mid 2
Cycle, Low
Cycle, Mid 1
Cycle, Mid 2
85%
20%
15%
10%
5%
80%
0%
ηCycle
ηCombust
35%
10%
20%
30%
40%
50%
EGR Rate (% Vol. of Intake Air)
0%
60%
Figure 43: Response of combustion and cycle efficiencies to EGR Rate, ethanol
HCCI combustion, 3 loads, 1500 RPM
7.2.2
Emissions Analysis
A summary of brake specific emissions from the low load EGR tests is shown in
Figure 41. It can be seen that brake specific NOx and PM both decrease as EGR is
increases, but CO and HC fall only slightly. The relatively small influence of EGR on
CO and HC emissions is due to the competing influences of temperature and oxygen
concentration. As EGR is increases, peak in-cylinder temperature generally rises as
shown in Table 10 but oxygen decreases due to increasing lambda as shown in Table 9.
Oxygen in the exhaust is reduced by nearly 60% at all loads from the 0% EGR
condition to the 50% EGR condition. Response of CO and HC emissions to EGR at mid
load 1 shown in Figure 44 is similar to low load.
100
70.0
0.016
Brake Specific CO, HC
(g/kW hr)
60.0
50.0
0.014
0.012
0.010
40.0
0.008
30.0
0.006
20.0
0.004
10.0
0.002
0.0
0.000
60%
0%
10%
20%
30%
40%
50%
EGR (% Vol. of Intake Air)
Brake Specific PM, NOX
(g/kW hr)
BSCO
BSHC
BSNOx
BSPM
Figure 44: Brake specific emissions from ethanol HCCI combustion with varying
EGR rate, 1500 RPM, 130°C intake temperature, low load
For all particulate size distributions presented here, confidence intervals are
established at the 90% level using a t-distribution, sample sizes vary by data set from
five to twenty. Error bars shown for BSPM in Figure 44, Figure 47, and Figure 50 also
represent a 90% confidence interval. Error bars shown for brake specific gas phase
emissions represent the standard error of the mean for each data point. Sample sizes
ranged from 2 to 7. Particle number and mass show significant sensitivity to EGR, with
mass levels spanning nearly 3 orders of magnitude. To clearly illustrate these variations
and the details of the individual distributions, particle mass data is shown on a log-log
plot.
Figure 45 and Figure 46 show size and mass distributions respectively for the low
load condition. Although little difference is seen in particulate emissions at low EGR
levels, as EGR reaches 25% and then 50%, a significant drop in both number and mass
is clear at each level. Taking into account the fixed fueling rate and relatively stable
cycle efficiencies, there are no detrimental effects of high EGR levels on engine
performance or emissions. Stable engine output coupled with the sharp drop in total
101
particulate mass leads to sharp drops in brake specific PM as EGR increases as shown
in Figure 44.
dN/dlogDP (part./cm3)
2.0E+08
EGR = 0%
EGR = 10%
EGR = 25%
EGR = 50%
1.5E+08
1.0E+08
5.0E+07
0.0E+00
1
10
100
DP (nm)
Figure 45: Mobility size distributions with varying EGR rate, ethanol HCCI
combustion, fixed fueling, 1500 RPM, 130° intake temperature, low load
For the low load and mid load 1 conditions, increasing EGR levels lead to
downward trends in both total number and particle mobility diameter. The number
concentrations for each of these loads are shown in Figure 45 and Figure 48
respectively. The combined effect of these two characteristics is a very significant
reduction in particle mass concentration. The corresponding mass distributions are
shown in Figure 46 and Figure 49.
The NOX - PM tradeoff, summarized by Ladommatos, et al. (1999), Abd-Alla
(2002), and Zheng et al. (2004), in traditional CI engines is clearly not present in these
HCCI data. This behavior represents an important advantage of the HCCI engine and is
consistent with overall findings in the HCCI literature.
102
1.0E-01
EGR = 0%
EGR = 10%
EGR = 25%
EGR = 50%
dM/dlogDP (µg/cm3)
1.0E-02
1.0E-03
1.0E-04
1.0E-05
1.0E-06
1.0E-07
1
10
100
DP (nm)
Figure 46: Mass distributions with varying EGR rate, ethanol HCCI combustion,
fixed fueling, 1500 RPM, 130° intake temperature, low load
0.040
BSCO
BSHC
BSNOx
BSPM
Brake Specific CO, HC
(g/kW hr)
35.0
30.0
0.035
0.030
25.0
0.025
20.0
0.020
15.0
0.015
10.0
0.010
5.0
0.005
0.0
0.000
60%
0%
10%
20%
30%
40%
50%
EGR (% Vol. of Intake Air)
Brake Specific PM, NOX
(g/kW hr)
40.0
Figure 47: Brake specific emissions from ethanol HCCI combustion with varying
EGR rate, 1500 RPM, 110°C intake temperature, mid load 1
103
dN/dlogDP (part./cm3)
4.0E+08
EGR = 0%
EGR = 10%
EGR = 25%
EGR = 50%
3.0E+08
2.0E+08
1.0E+08
0.0E+00
1
10
100
DP (nm)
Figure 48: Mobility size distributions with varying EGR rate, ethanol HCCI
combustion, fixed fueling, 1500 RPM, 110° C intake temperature, mid load 1
1.0E-01
EGR = 0%
EGR = 10%
EGR = 25%
EGR = 50%
dM/dlogDP (µg/cm3)
1.0E-02
1.0E-03
1.0E-04
1.0E-05
1.0E-06
1.0E-07
1
10
100
DP (nm)
Figure 49: Mass distributions with varying EGR rate, ethanol HCCI combustion,
fixed fueling, 1500 RPM, 110° C intake temperature, mid load 1
104
Figure 47 is a plot of brakes specific emissions against EGR for mid load 2. This
operating condition shows somewhat different emissions behavior from the lighter load
conditions. Both CO and HC emissions peak at 10% EGR and then fall, but CO rises
again after 25% EGR while HC continues to fall. The heat release data in Table 10
show that peak in-cylinder temperature initially falls between 0 and 10% EGR but then
increases. On the other hand, lambda and exhaust oxygen decrease steadily as EGR
increases. The opposing effects of in-cylinder oxygen and temperature may explain the
observed trend of CO but the opposite trends in CO and HC at the highest EGR rates
suggests that HC oxidation may be more temperature dependent that CO oxidation.
However another effect may be playing a role. Although peak in-cylinder temperatures
are at their highest, peak heat release rates fall significantly in the 50% EGR case. Due
to the slower rate of heat release for this condition, wall heat transfer may have more of
an opportunity to cool the mixture in the cylinder wall boundary layer leading to slow
oxidation in this region. However this does not explain the opposite trends in CO and
HC emissions at the highest EGR rates. In this scenario global cylinder temperatures
can remain high as the mixture combusts reasonably close to TDC. However as
indicated by the burn duration and HRR, the combustion process is relatively slow,
allowing the CAD timescale of combustion and timescale of heat transfer to approach
parity.
105
0.1
BSCO
BSHC
BSNOx
BSPM
Brake Specific CO, HC
(g/kW hr)
35
30
0.09
0.08
0.07
25
0.06
20
0.05
15
0.04
0.03
10
Brake Specific PM, NOX
(g/kW hr)
40
0.02
5
0.01
0
0%
10%
20%
30%
40%
50%
EGR (% Vol. of Intake Air)
0
60%
Figure 50: Brake specific emissions from ethanol HCCI combustion with varying
EGR rate, 1500 RPM, 100°C intake temperature*, mid load 2
dN/dlogDP (part./cm3)
2.0E+09
EGR = 0%
EGR = 10%
EGR = 25%
EGR = 50%
1.5E+09
1.0E+09
5.0E+08
0.0E+00
1
10
100
DP (nm)
Figure 51: Mobility size distributions with varying EGR rate, ethanol HCCI
combustion, fixed fueling, 1500 RPM, 100°C intake temperature*, mid load 2
106
1.0E-01
EGR = 0%
EGR = 10%
EGR = 25%
EGR = 50%
dM/dlogDP (µg/cm3)
1.0E-02
1.0E-03
1.0E-04
1.0E-05
1.0E-06
1.0E-07
1
10
100
DP (nm)
Figure 52: Mass distributions with varying EGR rate, ethanol HCCI combustion,
fixed fueling, 1500 RPM, 100°C intake temperature*, mid load 2
The trends in NOx and PM emissions shown in Figure 47 are somewhat more
complex than for the light load cases. Both show a local minimum between 10 and 25%
EGR followed by an increase and then a slow decrease. The trends in peak in-cylinder
temperature and peak HRR rate are also complex with a general increase in temperature
with EGR but with a dip in temperature a 10% EGR while HRR, like PM and NOx,
falls, rises, and then fall again as EGR increases. Although the relationship between incylinder temperatures and PM mass is not obvious, a clear relationship between HRR
and PM mass is apparent. Increased HRR caused total PM mass to trend upwards. This
represents similar behavior to that reported in SI engines by Kayes and Hochgreb
(1999), assuming higher loads are indicative of higher peak HRRs. Like the lighter load
conditions, a PM-NOx tradeoff with respect to EGR (Ladommatos, et al, 1999; AbdAlla, 2002; Zheng et al., 2004) is not present, the two emissions respond in the same
way to EGR, although the response of PM is stronger.
Just as in the variable intake temperature cases examined in Chapter 6 the particle
data are consistent with the view that particles emitted from pure HCCI combustion, at
least for the pure ethanol fuel used here, are mainly formed from partially burned
107
lubricating oil. Thus the findings of Tobias et al. (2001) and Sakurai et al. (2003) that
the volatile components of PM are very similar to lubricating oil may also apply here.
The difference is that unlike a Diesel engine there are no solid accumulation mode
particles to adsorb the oil so the resulting particles are nearly entirely volatile.
7.3 Conclusions
The effects of EGR on an ethanol fueled HCCI engine were studied at constant
speed and three loads. Data were collected on performance, in-cylinder behavior, and
emissions. At all loads the effect of increasing EGR leading to longer burn duration was
confirmed. This is in good agreement with multiple published findings. In most cases
studied, increases in EGR led to decreases in both NOx and total PM mass and number
emissions. The mid load 2 condition showed a slightly more complex trend with a
general trend of decreasing PM and NOx emissions with increasing EGR but local
minima for both pollutants between 10 and 25% EGR. This was a consequence of an
underperforming EGR cooler that led to more advanced combustion and elevated heat
release rates and in-cylinder temperatures at high EGR rates.
The influence of EGR on combustion behavior was as expected, extending burn
duration, limiting rates of pressure rise, and minimizing peak rates of heat release.
Cooler combustion led to small reductions in NOX as EGR rates were increased. CO
and HC emission remained relatively stable at each condition throughout varying EGR
rates most likely due to competition between increased cylinder temperatures promoting
oxidation and decreasing air to fuel ratios limiting oxygen available.
PM emissions appear to be formed from lubricating oil. In general, increased rates
of EGR led to lower PM number and mass concentrations and smaller particle
diameters. These reductions are thought to be due to the lower peak rates of heat release
leading to less heat transfer to cylinder walls and reductions in the rates of evaporation
of oil films from in-cylinder surfaces.
Conducting a multiple regression analysis on the entire ethanol HCCI with variable
EGR rate data set illustrates the dependence of total particulate mass on select
combustion parameters. The influence of the independent variables; SOC, combustion
108
duration, peak temperature, and peak heat release rate on the dependant variable, total
particulate mass, was analyzed. The size of the sample set used for analysis was 12,
which led an R2 value of .83.
109
Chapter 8
The Effects Fuel Blending on Emissions
in an Ethanol and Hydrogen Fueled HCCI Engine.
The advanced capabilities of onboard engine ECUs enable dual fuel technology to
be feasible for implementation in the transportation sector. Although the research
community has not agreed upon a scale to quantitatively gauge fuel ignition properties
in an HCCI mode, analogous to cetane in CI mode or octane in SI mode, there are
obvious fuel effects governing SOC. Utilizing an intake charge blended from two
distinct fuels gives a scenario of easily altered and precisely controlled global fuel
properties. This can enable the operator to manipulate charge properties easily by
varying the proportions of the two fuels. The interactions of fuel blending with
combustion timing are complex and vary significantly with the fuels selected. For this
study hydrogen and ethanol were chosen as the fuel pair to be studied because hydrogen
rich gases can be made relatively easily onboard by reforming ethanol. As reported by
Hosseini and Checkel (2007) different effects can dominate the interactions giving
hydrogen the ability to have directly opposite effects on ignition timing depending on
the base fuel. These effects are not the focus of this study which will be limited to the
influence of hydrogen addition on overall combustion timing and duration quantified
via combustion parameters; SOC, burn duration, peak heat release rate, peak
temperature, IMEP and emissions
The goal of this work is to look at fuel blending as a fast response means of
controlling combustion phasing and the corresponding influences on emissions. The
SOC advances reported by Yap et al. (2004, 2006), Hosseini and Checkel (2006, 2008)
altered timings on the order of 0 to 6 CAD. Although these are significant timing
changes, they do not offer adequate ignition enhancement to mitigate the need for
thermal conditioning of intake air. For the engine used in the current work operating the
engine in a HCCI mode still required some level of intake heating. The work described
in Chapter 6 explored optimization of engine output with intake temperature control.
These optimized intake temperature conditions are used as the starting point, or zero
hydrogen energy condition, for a series of fuel blending experiments. Changes in the
110
combustion process are quantified and the relationships between hydrogen energy
proportion and emissions characteristics are explored.
8.1 Experimental
Ethanol and hydrogen fuels were selected due to their viability as renewable fuels.
Ethanol, though controversial by source, is easily utilized through existing infrastructure
and is left with few barriers in terms of engine technology. Hydrogen as a primary fuel
faces many logistical and technological barriers. However on-board auto-thermal
reforming can provide the small amounts hydrogen rich gas from a variety of liquid fuel
sources. This series of experiments examines the effects seen in emissions as a range
ethanol and hydrogen proportions are tested. Testing was conducted at three load
conditions, with the previously optimized intake temperatures for neat EtOH HCCI used
in two of the three. At the Mid-Load 2 condition high rates of pressure rise, indicating
the onset of engine knock, were encountered with 25% H2 energy when the optimized
100°C intake temperature was used. The intake temperature was lowered to 95° to allow
a constant intake temperature to be maintained while only hydrogen energy proportion
was altered. A summary of the operating conditions is given in Table 11. Engine speed
was selected corresponding to the manufactures rated torque speed in order to ensure
smooth engine operating. Loads were selected at reasonable intervals representing low
to moderate engine loads. The highest load condition was bounded by the onset of
engine knock. Because of the different heating values of the two fuels, different blends
of ethanol and hydrogen require slightly different global fuel to air ratios to maintain
constant engine output.
111
Table 11: Fuel Blending Test Conditions
Engine Load Condition
Low
Mid 1
Mid 2
Intake
Temperature
(°C)
130
110
95
λ-Range
4.35-4.42
3.32-3.62
2.98-3.11
Load (Nm)
IMEP Range
(kPa)
% Hydrogen
Output Energy
Range
Flow Rate
Range of EtOH
Energy In
(kW)
Flow Rate
Range of H2
Energy In
(kW)
53
89
125
224-231
310-317
383-403
0-25
0-25
0-25
42.5-38.9
54.6-47.8
65.9-58.5
0-3.47
0-8.34
0-8.16
In order to maintain constant engine output while varying the H2:EtOH proportion,
the following procedure was used.
•
First, steady operation of the HCCI engine was achieved on neat ethanol
fuel.
•
Data was then collected at this condition to use as the 0% hydrogen energy
baseline.
•
The fuel injector pulse width was then shortened, reducing fuel flow, until
the desired percentage of torque, or output energy, was removed.
•
Last, hydrogen flow was turned on to the hydrogen injectors and the supply
pressure adjusted until the engine was again running at the initial output
torque.
112
8.2 Results and Discussion
The experimental results of supplementing ethanol HCCI combustion with
hydrogen fuel are presented below. Initially an analysis of combustion phenomena is
conducted, followed by an emissions analysis. The results of both are compared with
the literature and relationships between combustion properties and emissions in a dual
fuel HCCI engine are established.
8.2.1
Combustion Analysis
Figure 53 through Figure 55 show in-cylinder pressure behavior with the addition
of supplemental hydrogen to ethanol HCCI combustion. Each plotted data set was
computed via 40 cycle average and then again averaged across each of the 4 cylinders.
There is little change in the compression behavior because H2, N2, and O2, all diatomic
gases have essentially the same ratio of specific heats, γ. Ethanol is polyatomic and has
a lower value of γ but the fractional replacement of ethanol by hydrogen is small and
does not appreciably influence the overall value of γ during compression. Also shown in
Figure 53, Figure 54, and Figure 55 are plots of heat release rate versus crank angle
derived from the pressure data. In calculating HRR, a single zone model similar to that
presented by Stone (1999) is used. The cylinder contents are assumed to behave as ideal
gases composed of an initially specified fuel and air mixture. The rates reported are net
heat release rates and neglect heat transfer to the cylinder walls. Heat release analysis
allows quantitative calculation of SOC timing, defined by the crank angle at which 10%
of the heat energy of the fuel has been liberated (CA10) and burn duration CA90-CA10.
Additionally in-cylinder temperature is calculated from the heat release analysis.
At the low load condition the combustion effects of hydrogen were very modest;
however as loads increased, clear trends developed showing a distinct advance in SOC
timing with increased % hydrogen energy. These timing advances lead to higher peak
pressures and increased rates of pressure rise in all cases examined.
113
8000
220
0% H2
5% H2
10% H2
15% H2
20% H2
25% H2
Pressure (kPa)
6000
5000
180
140
100
4000
3000
60
HRR (J/CAD)
7000
2000
20
1000
0
-20
-30
-20
-10
0
10
Crank Angle (°ATDC)
20
30
Figure 53: In-cylinder pressure behavior of EtOH and H2 HCCI combustion,
varying H2 output power, 1500 RPM, low load, 130° intake temperature
300
0% H2
5% H2
10% H2
15% H2
20% H2
25% H2
7000
Pressure (kPa)
6000
5000
260
220
180
4000
140
3000
100
2000
60
1000
20
0
HRR (J/CAD)
8000
-20
-30
-20
-10
0
10
20
30
Crank Angle (°ATDC)
Figure 54: In-cylinder pressure behavior of EtOH and H2 HCCI combustion,
varying H2 output power, 1500 RPM, mid load 1, 110° intake temperature
114
340
0% H2
5% H2 300
10% H2
260
15% H2
20% H2 220
25% H2
180
7000
Pressure (kPa)
6000
5000
4000
140
3000
100
2000
60
1000
20
0
HRR (J/CAD)
8000
-20
-30
-20
-10
0
10
20
30
Crank Angle (°ATDC)
Figure 55: In-cylinder pressure behavior EtOH and H2 HCCI combustion, varying
H2 output power, 1500 RPM, mid load 2, 95° intake temperature
The general trends of the in-cylinder pressure behavior from ethanol HCCI with
supplemental hydrogen fueling agree well with those found during preliminary
modeling exercises. In both modeled and experimental work the effects of hydrogen
were modest but did serve to advance combustion. Increases in hydrogen energy
resulted in advancement of SOC. Differences between experimental and modeled
behavior due to heat loss to the cylinder walls is also evident here and show similar
trends as the intake temperature and EGR comparisons.
Exploring the in-cylinder pressure trends in Figure 53 to Figure 55 we see clear
agreement with the work of Yap et al. (2004) where increased hydrogen energy leads to
advances in combustion timing. We are however reporting more pronounced effects at
higher engine loads, differing from the findings of those authors. Yap et al. (2004) were
using natural gas as the primary fuel in an engine with compression ratios in the range
of 12 to 15:1. The IMEP range reported was from 200 to 350 kPa. This falls close to the
IMEP range of the current study which ranged from 220 to 400 kPa. Hosseini and
Checkel (2006) have presented similar work with natural gas fueled HCCI
115
supplemented by hydrogen rich reformer gas that also shows advances in SOC due to
hydrogen addition. The work of Hosseini and Checkel (2006) was done at IMEP values
ranging from 150 to 200 kPa with compression ratios ranging from 16.5 to 18:1.
A summary of combustion parameters is given in Table 12 for each of the three
loads tested. Start of combustion, quantified by CA10 is shown to advance by at most
3.5 crank angle degrees. Variability of the combustion data was shown to be low with
the standard error of the mean in peak pressure timing calculated across the four
cylinders of the engine ranging from .15 to .53 CAD. The coefficient of variation of the
IMEP data across the 4 cylinders ranged from 0.9 % to 4.1 %.
Table 12: Summary of combustion properties, ethanol HCCI with supplemental
hydrogen fueling, 1500 RPM, 3 loads
Burn
Peak
Peak
Hydrogen IMEP
SOC
Dur.
HRR
Temp.
Energy
(kPa) (ºATDC)
(CAD) (J/CAD)
(K)
-5
7
90
1530
0%
230
-5
7
100
1530
5%
230
-5
7
100
1530
Low
10%
230
Load
-5
6
100
1530
15%
230
-5
7
90
1530
20%
230
-5
6
100
1550
25%
230
-2
6
130
1580
0%
310
-3
6
130
1580
5%
320
Mid
-3
6
130
1590
10%
310
Load
-3
6
130
1600
15%
320
1
-3
5
140
1620
20%
320
-4
5
160
1650
25%
310
3
9
140
1600
0%
400
1
7
150
1590
5%
390
Mid
1
7
150
1610
10%
390
Load
0
6
160
1660
15%
400
2
0
6
160
1650
20%
390
-1
5
170
1700
25%
380
Note that as combustion is advanced in each case, peak HRR and peak in-cylinder
temperatures also increase. It can be seen that burn duration decreases as SOC
116
advances. The increases in peak cylinder temperatures result from very rapid
combustion occurring over a small interval near TDC. If we imagine discrete piston
movement near TDC and consider the consequences of liberating energy in a small
number of these discrete increments the relationship between burn duration peak HRR,
and peak temperature becomes more clear. Figure 56 shows plots of combustion
efficiency and cycle efficiency plotted against hydrogen energy fraction for the three
loads. Combustion efficiency increases with increasing hydrogen fraction. This is likely
related to the increases in peak temperatures with hydrogen addition. As peak cylinder
temperatures increase, more complete oxidation of the fuel to CO2 drives reductions in
CO and HC emissions, as manifested in increases ηCombustion. Increases in combustion
rates are also likely due to the chain branching step cited by Yap et al. (2004) associated
with the reaction of atomic hydrogen and diatomic oxygen to form O and OH radicals.
At each load cycle efficiencies are almost independent of hydrogen addition rate. This is
surprising because cycle efficiency is usually proportional to combustion efficiency. It
may be that increases in heat transfer shifts in combustion timing offset gains due to
increased combustion efficiency.
100%
Combust, Low
Combust, Mid 1
Combust, Mid 2
Cycle, Low
Cycle, Mid 1
Cycle, Mid 2
99%
98%
96%
45%
40%
35%
30%
ηCycle
ηCombust
97%
50%
95%
25%
94%
20%
93%
15%
92%
10%
91%
5%
90%
0%
40%
0%
10%
20%
30%
% Engine Output from Hydrogen
Figure 56: Response of combustion and cycle efficiencies to variations in H2:EtOH
proportion, dual fuel HCCI combustion, 3 loads, 1500 RPM
117
Combustion efficiency increases at each load as more hydrogen fuel is added.
Calculated in part from emissions of CO and HCs, combustion efficiency is a result of
lower emissions of these species and more complete oxidation of the fuel. These results
reflect the lower CO and HC emissions at higher hydrogen proportions shown in Figure
57, Figure 60, Figure 63. It should be noted however that hydrogen emissions
considered in these efficiencies are calculated from a carbon balance and not directly
measured.
8.2.2
Emissions Analysis
Figure 57 through Figure 64 give brake specific emissions along with detailed
particle mobility size distributions for the three load conditions tested. Confidence
intervals on the particle size and mass distributions were established using students tdistributions at the 90% confidence level. Sample size varied from five to ten depending
on the test condition. Error bars on the BSPM data are based on the same 90%
confidence interval. Errors bars shown for gas phase emissions data represent the
average standard error of the mean for the respective pollutant. These values were
compiled from data taken at all three loads.
Resulting from the minimal combustion changes seen in Figure 56 and Figure 53,
brake specific emissions shown in Figure 57 remain relatively stable. These results
show consistency with the stable combustion parameters calculated at the low load
condition in Table 12. Particle mass and number distributions also remain relatively
unchanged throughout the six hydrogen energy proportions at this load as shown in
Figure 58 and Figure 59.
118
0.1
BSCO
BSHC
BSNOx
BSPM
Brake Specific CO, HC
(g/kW hr)
100
0.08
80
0.06
60
0.04
40
Brake Specific PM,NOX
(g/kW hr)
120
0.02
20
0
0%
5%
10% 15% 20% 25%
H2 (% Output Power)
30%
0
35%
Figure 57: Brake specific emissions from EtOH and H2 HCCI combustion with
varying H2 energy, 1500 RPM, low load, 130°C intake temperature
3.5E+08
H2 = 0%
H2 = 5%
H2 = 10%
H2 = 15%
H2 = 20%
H2 = 25%
2.5E+08
3
dN/dlogDP (part./cm )
3.0E+08
2.0E+08
1.5E+08
1.0E+08
5.0E+07
0.0E+00
1
10
100
DP (nm)
Figure 58: Mobility size distributions from EtOH and H2 HCCI combustion with
varying H2 energy, 1500 RPM, low load, 130°C intake temperature
119
6.0E-03
H2 = 0%
H2 = 5%
H2 = 10%
H2 = 15%
H2 = 20%
H2 = 25%
dM/dlogDP (µg/cm3)
5.0E-03
4.0E-03
3.0E-03
2.0E-03
1.0E-03
0.0E+00
1
10
100
DP (nm)
Figure 59: Mass distributions from EtOH and H2 HCCI combustion with varying
H2 energy, 1500 RPM, low load, 130°C intake temperature
Brake specific emissions data from the mid load 1 condition are shown in Figure
60, and number and mass weighted particle size distributions are shown in Figure 61
and Figure 62, respectively. The combustion behavior summarized in Table 12 shows a
stronger response to hydrogen fraction that in the low load case with significant
increases in both peak temperatures and heat release rates with increasing hydrogen
fraction. In direct response to increased cylinder temperatures and heat release rates,
NOx emissions increase by nearly a factor of three and PM emissions by more than a
factor of five as hydrogen energy is increased from 0 to 25 percent. On the other hand,
CO and HC emissions fall as increasing cylinder temperatures promote more complete
oxidation of the fuel, which is also reflected in the combustion efficiency trends. In
addition, there is less fuel to form CO and HC as ethanol is replaced by hydrogen.
120
0.12
BSCO
BSHC
BSNOx
BSPM
Brake Specific CO, HC
(g/kW hr)
70
60
0.1
0.08
50
40
0.06
30
0.04
20
Brake Specific PM, NOX
(g/kW hr)
80
0.02
10
0
0%
5%
10% 15% 20% 25%
H2 (% Output Power)
30%
0
35%
Figure 60: Brake specific emissions from EtOH and H2 HCCI combustion with
varying H2 energy, 1500 RPM, mid load 1, 110°C intake temperature
Referring to Figure 61, particle number concentrations shift to a smaller peak
mobility diameter while increasing in concentration as hydrogen energy proportion is
increased. Although particle sizes are shifted to small diameters, the increases in
number concentrations drive mass concentrations up as well.
121
2.0E+09
H2 = 0%
H2 = 5%
H2 = 10%
H2 = 15%
H2 = 20%
H2 = 25%
dN/dlogDP (part./cm3)
1.5E+09
1.0E+09
5.0E+08
0.0E+00
1
10
100
DP (nm)
Figure 61: Mobility size distributions from EtOH and H2 HCCI combustion with
varying H2 energy, 1500 RPM, mid load 1, 110°C intake temperature
3.0E-02
H2 = 0%
H2 = 5%
H2 = 10%
H2 = 15%
H2 = 20%
H2 = 25%
dM/dlogDP (µg/cm3)
2.5E-02
2.0E-02
1.5E-02
1.0E-02
5.0E-03
0.0E+00
1
10
100
DP (nm)
Figure 62: Mass distributions from EtOH and H2 HCCI combustion with varying
H2 energy, 1500 RPM, mid load 1, 110°C intake temperature
122
The emissions for the mid load 2 condition are plotted in Figure 63. The general
trends are similar to those of mid load 1. Again an increase in brake specific NOX
emissions of roughly three fold is shown as peak in-cylinder temperatures climb by 100
K from the 0 to 25 percent hydrogen energy conditions. Expected behavior is shown in
CO and HC emissions and consequentially reflected in combustion efficiency.
Following similar trends as the mid load 1 condition, the mid load 2 tests show BSPM
increasing with increased hydrogen energy. However the increase in PM is much more
modest than in the previous case.
0.3
BSCO
BSHC
BSNOx
BSPM
90
Brake Specific CO, HC
(g/kW hr)
80
70
0.25
0.2
60
50
Brake Specific PM, NOX
(g/kW hr)
100
0.15
40
0.1
30
20
0.05
10
0
0%
5%
10% 15% 20% 25%
H2 (% Output Power)
30%
0
35%
Figure 63: Brake specific emissions from EtOH and H2 HCCI combustion with
varying H2 energy, 1500 RPM, mid load 2, 95°C intake temperature
Particle size and mass distributions are shown in Figure 64 and Figure 65.
Although a general trend towards decreased mobility diameter is shown, total PM mass
is still increased through elevated particle concentrations.
123
3.0E+09
H2 = 0%
H2 = 5%
H2 = 10%
H2 = 15%
H2 = 20%
H2 = 25%
dN/dlogDP (part./cm3)
2.5E+09
2.0E+09
1.5E+09
1.0E+09
5.0E+08
0.0E+00
1
10
100
DP (nm)
Figure 64: Mobility size distributions from EtOH and H2 HCCI combustion with
varying H2 energy, 1500 RPM, mid load 2, 95°C intake temperature
3.0E-02
H2 = 0%
H2 = 5%
H2 = 10%
H2 = 15%
H2 = 20%
H2 = 25%
dM/dlogDP (µg/cm3)
2.5E-02
2.0E-02
1.5E-02
1.0E-02
5.0E-03
0.0E+00
1
10
100
DP (nm)
Figure 65: Mass distributions from EtOH and H2 HCCI combustion with varying
H2 energy, 1500 RPM, mid load 2, 95°C intake temperature
124
Faster burn rates lead to elevated peak in-cylinder temperatures, which are directly
responsible for increases in NOX emissions due to the sensitivity of NO formation to
temperature. Additionally the higher peak temperatures associated with increased
hydrogen energy explain reductions in CO and HC emissions as a result of faster
oxidation.
These reductions of CO and HC emissions may also be in part due to displacement
of hydrocarbon energy with hydrogen energy. To separate the effects of hydrogen
enhancing ethanol combustion rather than hydrogen displacing the base fuel, a set of
ethanol fuel normalized CO and HC emissions parameters have been calculated in a
manner similar to Bika et al. (2009). Figure 66 shows CO and HC emissions normalized
to ethanol fueling rate. To further clarify the analysis each supplemental hydrogen
condition is normalized to the initial pure ethanol CO and HC emissions level. For each
of the three load conditions, a significant increase in oxidation of CO is shown with
increasing hydrogen energy. The trends are less pronounced for HC emissions, likely
due to the origination of hydrocarbons in the quench zone, where lower local
temperatures are less sensitive to changes in peak temperature.
Low CO
Mid 1 CO
Mid 2 CO
Low HC
Mid 1 HC
Mid 2 HC
Normalized CO or HC Rate
((PPM/gEtOH) / (PPM0/gEtOH))
1.25
1
0.75
0.5
0.25
0.0%
10.0%
20.0%
30.0%
% Hydrogen Energy
Figure 66: Ethanol fueling rate specific CO and HC emissions normalized with
respect to 0% hydrogen fueling
125
Conducting a multiple regression analysis on the entire ethanol with supplemental
hydrogen fueling data set brings to light a strong dependence of total particulate mass
on select combustion parameters. Again, the influence of the independent variables;
SOC, combustion duration, peak temperature, and peak heat release rate on the
dependant variable, total particulate mass, was analyzed. The size of the sample set used
for analysis was 18, which led an R2 value of .90. Comparing this value to the R2 values
obtained through the variable intake temperature and variable EGR data, we see the
strongest correlation here. A major contributor to the strength of this correlation is the
low variability of the data set. Hydrogen flowrates were easily set to each cylinder and
precisely controlled through the engine controller. Hydrogen flow was recorded
throughout all testing via a mass flow meter which measured to precision levels at
around .1% of the measured flow rate.
Looking in more detail at the relationship between combustion parameters and
emissions data, Figure 67 and Figure 68 illustrate the dependency of CO, NOX, and PM
emissions on in-cylinder temperatures and heat release rate. The relationship between
CO and NOX emissions and peak in-cylinder temperature behaves as expected as a
result of the dependency of reaction rate constants on temperatures, however the trend
exhibited by PM emissions is less clear. Nucleation mode particulate matter in CI
engines forms via homogeneous nucleation of sulfates and hydrocarbons during dilution
and cooling of the exhaust gas (Abdul-Khalek et al. , 2000; Kim et al., 2002;
Vaaraslahti et al., 2005; Ristimäki et al., 2007). Vaaraslahti et al. (2005) has suggested
that as fuel sulfur content diminishes, sulfur and hydrocarbons originating from the
lubricating oil become increasingly important as nucleation mode precursors. Gas to
particle conversions generate critical clusters, eventually growing via condensation to
measurable particles. Nucleation behavior, driven by saturation ratio (Kim et al., 2002)
would be significantly affected by the availability of additional precursor material in the
exhaust stream. Increased peak cylinder temperatures are indicative of higher overall
temperatures throughout the cycle, a condition that leads to elevated evaporation rates
of engine lubricating oil from atomized oil droplets and the cylinder walls (Yilmez et
al., 2002; Yilmez et al., 2004; Audeete & Wong, 1999; Gilles et al., 2007).
126
0.3
BSCO
BSPM
BSNOx
BSCO (g/kW hr)
100
R2 = 0.94
80
0.25
BSPM, BSNOX (g/kW hr)
120
0.2
60
0.15
R2 = 0.84
40
0.1
R2 = 0.93
20
0
1500
1550
1600
1650
1700
0.05
0
1800
1750
Peak in-cylinder temperature (K)
Figure 67: Brake specific emissions vs. peak in-cylinder temperature, ethanol
HCCI with 0 to 25% supplemental hydrogen fueling, 1500 RPM, 3 loads
120
0.3
BSCO
BSPM
BSNOx
0.25
R2 = 0.95
0.2
80
0.15
60
2
R = 0.85
0.1
40
20
R2 = 0.89
0
50
100
150
BSPM, BSNOX (g/kW hr)
BSCO (g/kW hr)
100
0.05
0
200
Peak HRR (J/CAD)
Figure 68: Brake specific emissions vs. peak heat release rate, ethanol HCCI with 0
to 25% supplemental hydrogen fueling, 1500 RPM, 3 loads
127
An early study conducted by Tobias et al. (2001) used thermal desorption particle
beam mass spectrometry (TDPBMS) and temperature programmed thermal desorption
(TPTD) to investigate Diesel particulate matter composition. Tobias et al. (2001)
conducted analysis of the volatile organic compounds comprising Diesel particulate
matter and found high percentages of alkanes and cycloalkanes. The similarity of the
cycloalkanes to alkanes ratios in the PM to those of the lubricating oil led the authors to
believe that significant contributions were made from lubricating oil. Sakurai et al.
(2003) further investigated Diesel nanoparticle composition through TDPBMS and
tandem differential mobility analyzer (TDMA) techniques and found volatile particles
emitted at low to moderate loads under steady state conditions were composed roughly
95% compounds from unburned lubricating oil. Additionally, an investigation by Miller
et al. (2007) into exhaust PM from pure hydrogen fueled SI engines noted that as incylinder temperature increased, organic carbon levels in the PM also increased. The
authors thought this was likely due to more complete breakdown and oxidation of some
lubricating oil components at elevated temperatures. This HCCI work suggests very
similar behavior is taking place here.
8.3 Pure Hydrogen HCCI
To further develop an understanding of the HCCI combustion process and its
emissions, an investigation was made into the behavior of pure hydrogen fueled HCCI
combustion. Hydrogen was selected due to its purity as a fuel. The fuel used for this
work was research grade high purity hydrogen; it was specified to have > 99.999 %
purity.
Recent work has demonstrated hydrogen as a viable HCCI fuel. Caton and Pruitt
(2009) explored hydrogen HCCI operating conditions from compression ratios of 17:1
up to 20:1 and were able to maintain controlled combustion at λ values ranging from
roughly 7 up to 2.5. Intake temperatures tested were in the rage of 80 ° to 100°C.
Operation was bounded on the high load end by advanced combustion leading to knock
and on the lower load end by drastic drops in efficiency. These efficiency losses are due
to inability of the combustion process to fully oxidize the fuel at the lowest loads tests.
128
Gomes – Antunes (2008) also explored hydrogen HCCI and noted that the high
rates of pressure rise found limited its use to relatively light load applications.
Compression ratios of 17:1 were used for this work and the engine ran successfully at λ
values ranging from 6 to 3. Intake temperatures ranged from 85° to 110°C.
The work of Tobias et al. (2001) and Sakurai et al. (2003) points towards
lubricating oil as the primary contributor of precursors to nucleation mode particles in
lean burning CI engines. Miller et al. (2007) has also drawn similar conclusions from SI
engines running on pure hydrogen fuel. Eliminating fuel bound hydrocarbons and
operating an HCCI engine on hydrogen will provide a means to further verify the
hypothesis that a similar case can be made for fully premixed HCCI engines. In
eliminating fuel bound impurities and hydrocarbons, the precursors to nucleation mode
PM are constrained and further speculation into precisely what compounds form this
PM in HCCI engines can be made.
8.3.1
Experimental
The same modified 5.2 liter Isuzu engine described in earlier experiments was used
here. For comparative purposes, the engine was operated with a load near 52 Nm at
1500 RPM, the same load and speed as all of the low load ethanol work. These
conditions also resulted in an IMEP range within that of the low load ethanol work. A
range of three intake temperatures were swept through to assess the effects intake
temperature variation. For all test cases hydrogen fueling was held constant.Table 13
gives the pertinent details of the experimental conditions.
Because conventional gas phase emission instrumentation concerned with HC, CO,
and CO2 are of little use when studying hydrogen fueled combustion, different exhaust
gas analysis equipment was used for the hydrogen HCCI work. A laser multi-gas
spectrometer (Atmosphere Recovery Inc.) was used to obtain concentrations of H2,
H2O, O2, and N2 in the exhaust stream. A chemiluminescence NOX analyzer (California
Analytical Instruments, 600 HCLD) was also used to collect NOX data during this work.
The temperatures of the two stage dilution system were maintained at 35°, 25°, and
35°C, for stage 1 dilution air, stage 2 dilution air, and the dilution tunnel water jacket
129
respectively. Dilution ratios were set to the average dilution ratios obtained during the
ethanol and hydrogen work; S1 = 15.6 and S2= 18.8. A SMPS (TSI 3080) comprised of
a nano DMA (TSI 3085) and ultra fine CPC (TSI 3025), with the sheath and aerosol
flows set to 15 and 1.5 lpm, was used for all particle measurements.
Table 13: Hydrogen fueled HCCI test conditions
Intake
Temperature
(°C)
Fueling Rate
(gH2/sec)
Fuel Input
Energy Rate
(kW)
λ
Load (N•m)
IMEP (kPa)
8.3.2
95°
100°
105°
.314
.314
.314
38.0
38.0
38.0
5.09
52
230
5.06
54
230
4.97
52
230
Combustion Analysis
The range of temperatures were selected in an effort to optimize IMEP for the
given engine load and speed condition. From Table 13 it can be seen that this occurs
with the intake temperature at 100 °C.
Figure 69 shows in-cylinder pressure data gathered at each intake temperature.
Consistent with the ethanol data, peak pressures and SOC both show a direct
relationship with intake temperature for the intake temperature range shown.
130
8000
220
T=95
T=100
T=105
Motoring
Pressure (kPa)
6000
180
140
5000
4000
100
3000
60
HRR (J/CAD)
7000
2000
20
1000
0
-20
-30
-20
-10
0
10
Crank Angle (°ATDC)
20
30
Figure 69: In-cylinder pressure traces of hydrogen HCCI combustion, fixed
fueling, λ = 5.08 - 4.97, 1500 RPM, varying intake temperature
Also shown in Figure 69 are plots of heat release rate versus crank angle derived
from the pressure data. In calculating HRR, a single zone model similar to that
presented by Stone (1999) is used. The cylinder contents are assumed to behave as ideal
gases composed of an initially specified fuel and air mixture. The rates reported are net
heat release rates and neglect heat transfer to the cylinder walls. Heat release analysis
allows quantitative calculation of SOC timing, defined by the crank angle at which 10%
of the heat energy of the fuel has been liberated (CA10) and burn duration CA90-CA10.
Additionally in-cylinder temperature is calculated from the heat release analysis. Table
14 shows pertinent combustion parameters for the hydrogen HCCI tests with varying
intake temperature. Variability of the combustion data was shown to be low with the
standard error of the mean in peak pressure timing calculated across the four cylinders
of the engine ranging from .18 to .4 CAD. The coefficient of variation of the IMEP data
across the 4 cylinders ranged from 1.8 % to 3.0 %. It can be clearly seen that at all
engine loads studied elevating intake temperatures leads to advances in SOC. More
advanced combustion in turn leads to higher peak heat release rates. With more heat
131
released prior to or very near TDC, the physical volume in which the energy is released
becomes smaller, and due to engine geometry, does not change as much per CAD. This
causes higher cylinder pressures and temperatures. IMEP values are very near those of
the low load ethanol testing allowing relevant comparisons between the two tests to be
made.
Table 14: Summary of combustion properties, hydrogen HCCI with varying
intake temperature, 1500 RPM, 54 Nm Load
Intake
Burn
Peak
Peak
IMEP
SOC
Temp.
Dur.
HRR
Temp.
(kPa) (ºATDC)
(°C)
(CAD) (J/CAD)
(K)
230
3
10
80
1300
95
Low
230
2
8
90
1330
100
Load
230
0
6
100
1390
105
SOC values, calculated via net heat release, increase from 3°ATDC to TDC as
intake temperatures are increased from 95° to 105°C. With this advanced combustion
comes higher peak heat releases rates, higher peak in-cylinder temperatures, and higher
peak pressures. Burn durations also become significantly shorter as intake temperatures
increase. In low temperature combustion processes like HCCI, peak cylinder
temperatures are kept relatively low through globally lean fuel and air mixtures. As a
consequence, combustion efficiency, which is dependent on partially oxidized fuel, can
be low. Thus action taken to increase peak combustion temperatures will lead to
increased combustion efficiency through more complete oxidation of the fuel.
Combustion and cycle efficiencies were calculated in a manner similar to that
described in 6.2.1 and are shown in Figure 70. A notable difference is the omission of
CO and HCs from the analysis. With hydrogen as the sole fuel, emissions of the carbon
based pollutants, CO and HCs, were well below the sensitivity of the exhaust gas
analyzers. PM emissions do however indicate that the exhaust stream is clearly not
completely void of hydrocarbons, however, on a mass basis they are negligible have no
effect on this analysis. Although cycle efficiencies remain relatively constant at these
temperatures, combustion efficiency exhibits a direct relationship, increasing more than
132
1 % as temperatures are raised from 95° to 105°C. The range of combustion efficiencies
found are in very good agreement with those reported by Caton and Pruitt (2009) for a
hydrogen HCCI engine with similar geometry.
100%
50%
Combust
Cycle
99%
40%
98%
35%
97%
30%
96%
25%
95%
20%
ηCycle
ηCombust
45%
15%
94%
10%
93%
5%
92%
0%
110
90
95
100
105
Intake Temperature (°C)
Figure 70: Effect of intake temperature on combustion and cycle efficiency,
hydrogen HCCI combustion, fixed fueling, λ = 5.08 - 4.97, 1500 RPM
8.3.3
Emissions Analysis
Figure 71 summarizes the brake specific emissions and their response to variations
in intake temperature of the hydrogen fueled HCCI engine. BSNOX, BSH2, and BSPM
are given at each temperature. Increasing intake temperature leads to increased BSPM
emissions and decreased BSH2 emissions. However, BSNOX emissions remain
relatively stable. Decreases in hydrogen emissions, and the resulting increases in
combustion efficiency, are caused by more advanced combustion leading to higher incylinder temperatures, which promote more complete burning of the fuel. Advanced
combustion and high in-cylinder temperatures lead to higher PM emissions, showing
good agreement with earlier findings in Sections 6.2, 7.2, and 8.2. Full particle size and
133
mass distributions are given in Figure 72 and Figure 73 respectively. Particle size and
mass distributions obtained when motoring the engine are also presented. The motored
engine was operating hot with 120°C intake air, resulting in 80°C exhaust, and 96°C oil
temperatures.
Comparing hot motored emissions from the engine enables exploration into the role
of lubricating oil in HCCI emissions. Total mass concentration of motored PM
emissions from the HCCI engine is roughly 4600 µg/m3. This is more than two times
the highest levels present during the hydrogen HCCI testing. The primary reason for
elevated particulate emissions during motoring, which stems from high oil
consumption, is the inability of the piston rings to seat properly during the sharp
pressure drop during motored expansion. Furthermore temperatures are still high
enough to evaporate hydrocarbons from the lubricating oil to form nucleation precursor
material.
A simple polytropic compression model like that presented in Equation 7 gives an
idea of the range of temperatures encountered during motored compression in this
engine. Using a compression ratio of 18.5, an initial temperature of 120°C, and a γ value
of 1.3, gives peak cylinder temperatures around 950°C. Temperatures calculated for this
engine through in-cylinder pressures analysis are also very close to that. This indicates
that throughout a significant portion of the compression and expansion strokes during
motoring, gas temperatures are well above those required for evaporation of engine oil.
An additional contributor to the significantly elevated hot motored PM emissions is
the inability of the sub 1000°C motored cylinder temperatures to oxidize any of the
evaporated hydrocarbons. This results in all of the evaporated volatile material from the
lubricating oil being available as precursor to nucleation.
134
0.1
BSH2
BSNOx
BSPM
9
8
0.09
0.08
7
0.07
6
0.06
5
0.05
4
0.04
3
0.03
2
0.02
1
0.01
0
90
95
100
105
Brake Specific PM, NOX
(g/kW hr)
Brake Specific Unburned H2 (g/kW
hr)
10
0
110
Intake Temperature (°C)
Figure 71: Brake specific emissions from hydrogen HCCI with varying intake
temperature, fixed fueling, λ = 5.08 - 4.97, 1500 RPM
From Figure 72 is evident that increased intake temperatures lead to higher number
concentrations of particulate matter and larger mobility diameter particles. These
increases both contribute to increases in total mass, as illustrated by BSPM emissions in
Figure 71, and PM mass distributions shown in Figure 73. Error bars for the BSPM data
shown in Figure 71 represent 90% confidence intervals. For the BSNOX data, error bars
shown represent the average standard error of the mean for data taken during the EtOH
and H2 fuel blending experiments.
135
1.0E+09
H2, Tin = 95
H2, Tin = 100
H2, Tin = 105
Motoring
dN/dlogDP (part./cm3)
8.0E+08
6.0E+08
4.0E+08
2.0E+08
0.0E+00
1
10
DP (nm)
100
Figure 72: Mobility size distributions from a hydrogen fueled HCCI engine,
λ = 5.08 - 4.97, 1500 RPM, varying intake temperature
1.5E-02
H2, Tin = 95
H2, Tin = 100
H2, Tin = 105
Motoring
dM/dlogDP (µg/cm3)
1.2E-02
9.0E-03
6.0E-03
3.0E-03
0.0E+00
1
10
DP (nm)
100
Figure 73: Mass distributions from a hydrogen fueled HCCI engine,
λ = 5.08 - 4.97, 1500 RPM, varying intake temperature
136
Conducting two single linear regression analyses on the hydrogen HCCI with
variable thermal conditioning data set illustrates the dependence of total particulate
mass on select combustion parameters. Because of the limited number of data points,
and previous analyses pointing to peak temperature and peak HRR as the most
significant drivers of PM formation, the influence of these two variables on the
dependant variable, total particulate mass, were analyzed separately. The size of the
sample set used for the analysis was three. These data are shown in Figure 74 with peak
in-cylinder temperature shown on the lower axis and peak HRR shown on the upper
axis. Increased intake temperatures lead to more advanced combustion giving higher incylinder temperatures and higher rates of heat release with hydrogen HCCI in a similar
manner as in ethanol HCCI combustion. Also similar to ethanol HCCI combustion, PM
emissions rates from hydrogen HCCI increase with increasing in-cylinder temperatures
and peak HRRs.
Peak Heat Release Rate (J/CAD)
85
90
95
100
0.08
BSPM (g/kW hr)
0.07
Temp
Peak HRR
0.06
0.05
0.04
0.03
0.02
0.01
0
1200
R2 = 0.9766
R2 = 0.8377
1250
1300
1350
1400
Peak Temperature (K)
Figure 74: BSPM vs. peak HRR or peak temperature, neat hydrogen HCCI, 1500
RPM, low load, 3 intake temperatures
137
If we recall that the intake temperature for the 0% hydrogen energy condition in the
ethanol with supplemental hydrogen fueling tests was found by optimizing output
torque with intake temperature in earlier experiments, a useful analysis can be
developed by comparing those results to the intake temperature optimized peak torque
condition using pure hydrogen fuel for HCCI combustion found here. For hydrogen
HCCI the peak output condition, indicated by both torque and IMEP, corresponds to a
100°C intake temperature. The load, speed, and IMEP are the same for both the ethanol
and pure hydrogen cases. From Figure 75 it can be seen that the number and mass
concentration of particulate emissions are also strikingly similar. This insensitivity of
particulate emissions to fuel, especially when a hydrocarbon free fuel such hydrogen is
used, suggests the primary contributor to PM emissions in fully premixed HCCI
combustion are hydrocarbons evaporated from atomized lubricating oil droplets and the
cylinder wall. Miller et al. (2007) has made a similar conclusion for a hydrogen fueled
SI engine, noting that formation of primary soot particles, composed of elemental
carbon, generally takes place in a flame where carbon containing fuel is burned in a
locally oxygen starved environment. Miller et al. also presents the argument that
organic carbon emissions, found to be increasingly present at elevated loads during
hydrogen combustion, result from more complete breakdown of the lubrication oil at
higher in-cylinder temperatures. They also cited possible impurities in the fuel and
sources of seed particles for nucleation. In an attempt to minimize any fuel
contributions to PM formation, >99.999% pure hydrogen was used for the HCCI testing
done here.
138
6.0E-03
Neat Ethanol
Neat Hydrogen
dM/dlogDP (µg/cm3)
5.0E-03
4.0E-03
3.0E-03
2.0E-03
1.0E-03
0.0E+00
1
10
100
DP (nm)
Figure 75: Neat ethanol and neat hydrogen mass distributions, HCCI combustion,
1500 RPM, Load ≈ 54 Nm, IMEP ≈ 230 kPa, λEtOH = 4.4, λH2 = 5.0
8.4 Conclusions
A study on the combustion and emissions effects of supplemental hydrogen fueling
in a HCCI engine using ethanol as the primary fuel was conducted. The results clearly
indicated that combustion phasing is advanced with increasing hydrogen energy
proportion. Additionally, the effect became more pronounced at higher engine loads.
The advances in combustion phasing were in agreement with published findings for
another high octane fuel, natural gas.
Examining emission trends between loads, increased load generally led to higher
BSNOX and BSPM emissions. Within every load condition, a general trend of increases
in hydrogen energy proportion giving increased BSNOX and BSPM emissions was
present. PM present at all loads was composed entirely of nucleation mode particles,
with number concentrations virtually nonexistent above 60 nm in mobility diameter.
Significant reductions of ethanol fuel specific CO emissions were reported at all
loads as more hydrogen energy was added. These reductions, which measured more
than 50% at moderate loads with the highest hydrogen fueling rates, have been
139
attributed to enhanced oxidation of CO due to radicals produced via hydrogen chain
branching reactions. Modest reductions in ethanol fuel specific HC emissions were also
observed.
The sensitivity shown of PM to cylinder temperatures and HRR indicate the
primary source of precursor material is likely lubricating oil evaporated from atomized
droplets created by reverse gas flow through the ring pack, or evaporated directly from
cylinder walls during expansion. Because particulate matter is solely present in the
nucleation mode it is likely not originating from the soot forming regions found within
locally fuel rich combustion flames.
Additionally a similar engine condition fueled by neat ethanol or neat hydrogen
yielded nearly identical particulate emissions. These results strongly suggest lubricating
oil as the primary, if not sole contributor, to nucleation mode particulate matter in fully
premixed ethanol HCCI combustion. In both ethanol fueling supplemented with
hydrogen and pure hydrogen fueled HCCI combustion, emissions of NOX and PM
increased with increases in peak heat release rate and peak in-cylinder temperature.
Associated with the increased in-cylinder temperatures were shorter burn durations,
decreasing CO and HC emissions, and increasing combustion efficiency.
140
Chapter 9
Advanced Characterization Techniques
for Emissions from an Ethanol Fueled HCCI Engine
After establishing in previous chapters that particulate emissions in fully premixed
HCCI combustion are primarily formed from unburned lubricating oil, it becomes
necessary to further elucidate details of what components of the oil eventually reside in
exhaust PM. Utilizing TDMA techniques with a thermal conditioning section allow
particle volatility to be investigated. Additionally a much more detailed map of gas
phase emissions has been drawn through FTIR analysis of ethanol HCCI emissions.
9.1 TDMA Experiments
To gain additional insight into the composition of the particles formed during
HCCI combustion, a set of TDMA experiments were conducted at four engine loads
and a fixed engine speed of 1500 RPM. The loads corresponded to the low load, mid
load 1, mid load 2, and motoring conditions described in earlier experiments.
Taking the pioneering TDMA work of Liu, et al. (1978), McMurry, et al. (1983),
and Rader et al. (1986), and applying it to engine exhaust particles, has enabled
researchers to gain a great deal of insight into their composition. Orsini (1998) and
Sakurai (2003) have used TDMA techniques with thermal conditioning to study Diesel
nanoparticle composition and volatility.
Recently, Surawski et al. (2010) have utilized TDMA techniques to investigate the
details of particle composition in a Diesel engine with supplemental ethanol fumigation.
At all loads tested they found the addition of ethanol greatly increased the volatile
volume fraction of accumulation mode particulate matter in a compression ignition
engine. The increases in volatile volume fraction also followed directly as higher
percentages of ethanol energy were substituted.
Sharp increases in nucleation mode particles with increasing ethanol are explained
by a deficit in adsorption sites created through the mitigation of the accumulation mode.
Kittelson et al. (2002) explains the role of accumulation mode particles, composed of
carbonaceous agglomerates, as that of a condensation and adsorption sink for organic
141
vapors. When clean burning combustion processes lead to reduction of the
accumulation mode, these supersaturated vapors are left as precursors for nucleation of
new particulate matter.
9.1.1
Experimental
To isolate volatility fractions of the PM sampled from the engine, the exhaust
aerosol was studied with a TDMA apparatus that utilized a thermal conditioning section
(Orsini, 1998; Sakurai et al., 2003; Surawski et al., 2010). A schematic of the apparatus
is shown in Figure 11. The setup employed a long DMA (TSI 3081) as the fixed voltage
DMA which is used to size select particles prior to the thermal conditioning section.
Engine exhaust was initially sampled at each load condition with the fixed voltage
(long) DMA bypassed and the thermal conditioning section at ambient temperatures to
collect full particle size distributions. After determining the mode of the particle size
distribution from this data, the aerosol path was then routed through the fixed voltage
DMA, with its voltage set to correspond to the mode of the initial size distribution. At
each of the three fired engine loads, and a fourth motored load, particle size distribution
modes were found and the fixed voltage DMA set in a similar manner.
The nano DMA (TSI 3085) was operating downstream of the thermal condition
section in combination with a CPC having a D50 of 3.0 nm (TSI 3025), these two
instruments were operated as an SMPS and used to collect a size distribution of the
initially classified aerosol after being passed through the thermal conditioning section as
shown in Figure 11. The methodology allows a particle size change in the monodisperse
aerosol due to evaporation of volatile material from the particle surfaces in the thermal
conditioning section to be documented.
Both DMA columns were operated with a sample flow of 1.5 lpm and a sheath flow
of 15 lpm. At this flow rate, residence time in the thermal conditioning section was
about .25 seconds. The thermal conditioner was sized to give residence times in
accordance with Orsini (1998) and Sakurai (2003), with temperature monitored
continuously at the conditioner outlet.
Samples were then taken with thermal conditioning temperatures ranging from 40°
to 110° C in 10° increments. The thermal conditioning section was given adequate time
142
to stabilize at each increment and monitored with a type K thermocouple throughout the
testing.
Engine operating parameters of the three fired loads corresponded to the peak
IMEP intake temperature conditions which were isolated in earlier experiments and are
summarized in Table 15. The operating details of the motoring load condition are also
shown. For the fired loads the engine was operating on ethanol with the fuel content
specifications listed in Table 6. In order to maintain as high of particle counts as
possible the exhaust aerosol was sampled after one stage of dilution. The dilution ratio
was held at 15.6:1 throughout all testing. Stage one dilution air temperature was held at
35°C, as was the dilution tunnel water jacket.
Table 15: Engine operating parameters tested in TDMA analysis of ethanol HCCI
combustion
Speed
Load
IMEP
Intake
Condition
λ
(RPM)
(Nm)
(kPa)
Temp. (°C)
1500
120
Motoring
1500
53
230
130
4.3
Low Load
1500
89
320
110
3.5
Mid Load 1
1500
128
400
100
3.0
Mid Load 2
Because two different model DMAs were used, subtle differences in DMA
geometry, flow rate and power supply performance could lead to differences in overall
DMA performance. The following exercise was conducted to gauge how closely the
classifying performance of the long DMA agreed with that of the nano DMA. First, a
polydisperse dioctylsebacate (DOS) in isopropyl alcohol aerosol was generated with a
Collison atomizer. The aerosol was then diluted and passed through a diffusion dryer
containing activated carbon, leaving a pure DOS aerosol. The pure DOS particle size
distribution had a mode diameter of 60 nm with total concentrations near 2.5 x 106
particles/cm3. Next the fixed voltage (long) DMA was set to classify a particle size from
the polydisperse DOS aerosol. Finally the scanning voltage (nano) DMA was used to
report a measured mobility diameter. Diameters of 25 nm and 35 nm were selected as
modes representative of the ethanol HCCI particulate matter. The reported modes are
143
given in Table 16, an average offset correction of + 0.9 nm has been calculated and will
be applied to all TDMA data. The average offset is computed from three SMPS scans at
each of the two selected particle diameters.
Table 16: TDMA bias error data
Classified
Mobility
Diameter
(nm)
25
35
Thermal Conditioning Setpoint
T = 30°
T = 40°
Measured
Measured
Offset
Mobility
Offset
Mobility
(nm)
Diameter
(nm)
Diameter
(nm)
(nm)
23.3
-1.7
24.1
-0.9
34.6
-0.4
34.2
-0.8
Engine exhaust was initially sampled at each load condition with the fixed voltage
(long) DMA bypassed and the thermal conditioning section at ambient temperatures to
collect full particle mobility size distributions. The aerosol path was then routed through
the fixed voltage DMA with its voltage set to correspond to the mode of the initial
mobility size distribution. At each of the four engine loads a new mode was found and
the fixed voltage DMA set in a similar manner.
Samples were then taken with thermal conditioning temperatures ranging from 40°
to 110° C in 10° increments. The thermal conditioning section was given adequate time
to stabilize at each increment and monitored with a type K thermocouple throughout the
testing. The setup and flow paths of the TDMA apparatus are illustrated in Figure 11.
9.1.2
Results and Discussion
The four data sets collected are shown below, each has the initial full particle size
distribution shown for reference along with the TDMA data collected at every
temperature increment. The data shown is corrected for a 0.9 nm offset between the
long and nano DMAs. The right hand scale of each graph corresponds to the full
distributions, with the TMDA classified concentrations read from the left hand scale.
144
8.0E+06
dN/dlog DP (part./cm3)
6.0E+08
T=40 C
T=50 C
T=60 C
T=70 C
T=80 C
T=90 C
T=100 C
T=110 C
Full Dist.
6.0E+06
4.0E+06
5.0E+08
4.0E+08
3.0E+08
2.0E+08
2.0E+06
dN/dlog DP (part./cm3)
1.0E+07
1.0E+08
0.0E+00
1
10
DP (nm)
0.0E+00
100
Figure 76: Full distribution and TDMA data, motoring load, 1500 RPM
5.0E+06
dN/dlog DP (part./cm3)
1.6E+08
T=40 C
T=50 C
T=60 C
T=70 C
T=80 C
T=90 C
T=100 C
T=110 C
Full Dist.
4.0E+06
3.0E+06
2.0E+06
1.4E+08
1.2E+08
1.0E+08
8.0E+07
6.0E+07
4.0E+07
1.0E+06
dN/dlog DP (part./cm3)
6.0E+06
2.0E+07
0.0E+00
1
10
DP (nm)
0.0E+00
100
Figure 77: Full distribution and TDMA data, low load, 1500 RPM
145
5.0E+06
dN/dlog DP (part./cm3)
1.6E+08
T=40 C
T=50 C
T=60 C
T=70 C
T=80 C
T=90 C
T=100 C
T=110 C
Full Dist.
4.0E+06
3.0E+06
2.0E+06
1.4E+08
1.2E+08
1.0E+08
8.0E+07
6.0E+07
4.0E+07
1.0E+06
dN/dlog DP (part./cm3)
6.0E+06
2.0E+07
0.0E+00
1
10
DP (nm)
0.0E+00
100
Figure 78: Full distribution and TMDA data, mid load 1, 1500 RPM
2.0E+07
7.0E+08
T=40 C
T=50 C
T=60 C
T=70 C
T=80 C
T=90 C
T=100 C
T=110 C
Full Dist.
1.2E+07
8.0E+06
6.0E+08
5.0E+08
4.0E+08
3.0E+08
2.0E+08
dN/dlog DP (part./cm3)
dN/dlog DP (part./cm3)
1.6E+07
4.0E+06
1.0E+08
0.0E+00
1
10
DP (nm)
0.0E+00
100
Figure 79: Full distribution and TMDA data, mid load 2, 1500 RPM
146
From Figure 77, Figure 78, and Figure 79, it is clear that each of the three fired
load data sets show distinct increases in particle concentrations at the 90°, 100°, and
110°C TDMA temperature settings. A possible explanation lies in the thermodynamics
of the system downstream of the thermal conditioning section. As a result of the
elevated temperatures of this section of the TDMA, volatile materials are evaporated
from the surface of the particles with more material evaporated as temperatures
increase. At the highest temperatures, 90°, 100°, and 110°C, very few particles remain
to adsorb the condensable volatile matter that has been evaporated into the gas stream.
With concentrations of volatile material increasing and no available sites for adsorption,
homogeneous nucleation is becomes likely to occur at the exit of the thermal condition
section where temperatures abruptly fall creating a region of super saturation.
Figure 80 shows the change in mode versus TDMA thermal conditioning
temperature for the data collected. It can be seen that data from the fired engine loads
exhibits similar evaporative behavior. For each of these three cases, the primary
changes in mobility diameter occur below 90°C with a very similar trend shown for
each load. Data from the motoring condition however, exhibits a very different trend.
Particle growth here exhibits a negative, nearly linear trend. Change in particle diameter
for droplets composed of pure compounds C28, C30, and C32 are also shown. Growth
rates for these hydrocarbons were calculated using Equation 25. In this equation taken
from Hinds (1999), the growth rate is expressed in terms of molecular weight (MW), a
condensation coefficient (αC), vapor pressure of the liquid (pd), density of the liquid (ρ−
p),
Avogadro’s constant (NA), single molecule mass (m), Boltzmann’s constant (k), and
temperature (T). The equation is valid for particles with diameter smaller than mean free
path.
d (d p )
dt
=
2MWα C (− p d )
ρ p N A 2πmkT
25
Sakurai et al. (2003) modeled evaporation profiles of pure hydrocarbons C24, C28,
and C32 and found similar trends of PM evaporation profiles closely agreeing with pure
hydrocarbon droplets.
147
0
-5
∆ DP (nm)
-10
Motoring
Low Load
Mid Load 1
Mid Load 2
C28
C30
C32
-15
-20
-25
-30
30
40
50
60
70
80
90
100
110
120
Temperature (°C)
Figure 80: Evaporation profiles particulate matter from an ethanol fueled HCCI
engine at three fired loads and a motored load, 1500 RPM
The distinct difference in the motored and fired evaporation profiles leads to the
hypothesis that some of the less volatile components in the lubricating oil are either
burned or broken down to more volatile compounds during combustion. Differences in
the total diameter change of fired load data are simply a result of differing initial
particle diameter and a constant percentage of particle volume boiled off. The leveling
off behavior exhibited at the three fired loads is an artifact of the droplet being fully
evaporated which means no further loss in diameter can occur.
To further illustrate how these profiles affect total volume of PM, Figure 81
presents the remaining particle volume as a fraction of the initial particle volume found
at 40°C. Due to laboratory conditions, this temperature is the minimum temperature that
could be stably maintained by the PID controlled heater.
148
Remaining Volume of Total PM (%)
100%
Motoring
Low Load
Mid Load 1
Mid Load 2
80%
60%
40%
20%
0%
30
40
50
60
70
80
90
100
110
120
Temperature (°C)
Figure 81: Remaining volume fraction of PM in ethanol HCCI exhaust after
thermal conditioning during TDMA analysis, 4 loads, 1500RPM
It is shown in Figure 81 that roughly 98% of particle volume (or mass) is composed
of volatile material at fired engine loads. Contrasting these results with motored
operation, we see a much higher fraction of low volatility material. The volume fraction
curve for the motored operating indicates that nearly 20% of this particulate matter is
composed of much less volatile components.
Previous work by Tobias et al. (2001), Sakurai et al. (2003), and Vaaraslahti et al.
(2005) examining the origins of PM in compression ignition engines has indicated a
significant contribution from lubricating oil in the formation of high volatility exhaust
aerosols. Comparing the data in Figure 81 with the results presented by Sakurai et al.
(2003), very similar behavior can be seen. Although a completely different fuel, engine,
and combustion mode were used in the Sakurai study, the trends in remaining volume
fraction of PM closely agree with those presented here. In both cases roughly 90% of
particle volume is evaporated between 50° and 100°C.
After showing that nearly all of the particulate matter from this combustion process
is composed of volatile components likely originating from lubricating oil, the
149
following analysis is intended to aid in understanding the mechanism by which volatile
species are drawn from lubricating oil on the piston walls, top land area, and ring pack.
Figure 82 illustrates the piston and cylinder wall interface. For our purposes, the
combustion zone will be assumed to exist as a homogeneous single temperature zone
through which heat is added to the system. The boundaries are shown in red with the oil
film on the cylinder liner shown in green. Note that the oil film does not extend
completely around the inside of the cylinder liner. Because the film is dispersed each
stroke by the piston motion, the upper most regions of the cylinder do not receive a
fresh oil coating and thus provide no film surface area for evaporative transport. Shin et
al. (1983) list oil film thicknesses between the top piston ring and the cylinder wall on
the order of 1 to 8 µm, with a dependence shown on piston position.
Cylinder Liner
Combustion Zone
Detail Area
Piston
Oil Film
Detail Area
Figure 82: Fuel and air charge, piston, and cylinder liner interface
A recent investigation by Yilmaz et al. (2004) into engine oil consumption has
defined five possible routes of oil consumption in SI engines. The first, throw off, is due
150
to mechanical transport of oil from the top land and upper compression ring caused by
the inertial forces of the reciprocating piston. The second, transport via reverse gas
flow, is caused by a reversed pressure difference between the combustion chamber and
trapped volume between piston rings during expansion. In this mechanism of oil
consumption oil is initially driven past the first piston ring by high combustion
pressures. During expansion, pressure in the combustion chamber quickly falls while
pressure behind the first ring remains high, creating a pressure difference which forces
combustion gas and entrained oil through the small ring crevices. The third route,
known as blow by, becomes significant when crankcase fumes are directed back into
the intake manifold, bringing entrained oil droplets with the air flow. The fourth route is
evaporative mass transport of oil from the piston and cylinder liner. And the final route
is bulk mass transport via valve leakage between the valve and valve guide. However
the authors state this route has essentially eliminated by modern valve seals. A more
detailed examination of evaporative mass transfer from the piston and cylinder liner by
Yilmaz et al. (2002) suggested that it was the primary contributor to total engine oil
consumption during normal engine operation.
Of these routes, throw off, reverse gas flow, and evaporative mass transfer are
likely active contributors during both motoring and firing engine operation of the HCCI
engine studied here. A simple model of polytropic compression gives an idea of the
range of temperatures encountered during motored compression in this engine. Using
the engines compression ratio of 18.5, an initial temperature of 120°C, and a γ value of
1.3, gives peak cylinder temperatures around 950 °C. This would indicate that
throughout a significant portion of the compression and expansion strokes gas
temperatures are well above engine oil and coolant temperatures, thus ensuring heat
flow to the cylinder wall. This heat flow causes significant evaporation of oil from the
cylinder liner oil film. Furthermore, motoring pressures were measured near 5 MPa,
high enough to providing a significant pressure gradient to drive reverse gas flow
processes. Additionally, oil throw off, a function of inertial effects due to piston motion
contributes to atomization of oil during motoring operation. The engine was motored at
151
1500 RPM, the same speed as the fired conditions giving a similar set of driving forces
for oil throw off in all tests.
With higher in-cylinder temperatures expected during the combustion processes
associated with the fired loads, the reduced particulate emissions as compared with the
motored load seems initially counterintuitive. However these results could be partially
attributed to breakdown and oxidation of the volatile material evaporated from the oil as
part of the combustion process. Additionally, Yilmez et al. (2004) reports oil transport
via reverse gas flow through the ring pack is high at idle and drops as load is increased
to roughly 50% load where it stabilizes. The increase in reverse gas flow is attributed to
a decrease in blow by as load is decreased which leaves more oil available on piston
ring and top land surfaces for transport into the cylinder via reverse gas flow. Yilmez
also reports total oil consumption rates to reach a minima at about 50% load due to
competing effects from evaporation, blow by, throw off, and reverse gas flow.
Furthermore, piston rings are generally designed to seat best when operating at high
engine loads, thus high in-cylinder pressures. At very low loads, such at motoring, ring
seating is likely at its worst.
Particle size distributions collected with and without a catalytic stripper are shown
in Figure 83. Data is shown for a representative ethanol HCCI condition and a motoring
condition on the same engine. The details of the engine operating conditions are listed
in Table 15. The data is intended to give further insight into the differences between the
volatile fractions of particles generated during fired and motored engine processes.
Looking at the motored data, a clear distribution can be observed in the catalytic
stripper data with a mode near 7 nm and a total concentration of 2.81 x 107
particles/cm3. Without the catalytic stripper, the motored size distribution has as mode
near 20 nm with total concentrations near 4 x 108 particles/cm3. Apple et al. has shown
lubricating oil to have roughly 1.0 % ash content from trace metals, which is likely the
solid core evident in our data as well. Apple also showed that particle diameters above
30 nm contribution disproportionately more to total ash concentrations than smaller
particles. With motoring distributions showing significantly higher numbers of large
particles than fired distributions, the work of Apple et al. provides a good explanation
152
for why particles from motored loads have a significant solid fraction and those from
fired loads do not. The total PM mass concentrations from motoring operating are
nearly 3300 µg/m3 with solid ash contributing roughly 0.1% of that. The significant
levels of solid residue are a result of high oil consumption during motoring. From the
Mid Load 1 size distribution collected with the catalytic stripper, it is clear that there
were too few particles to develop a size distribution. Using the catalytic stripper to
collected data from the Low Load and Mid Load 2 engine conditions gave similar
results.
1.0E+09
3
dN/dlogDp (#/cm )
1.0E+08
1.0E+07
Motored
Motored, CS
Mid Load 1
Mid Load 1, CS
1.0E+06
1.0E+05
1
10
100
DP (nm)
Figure 83: Particle size distributions collected with and without a catalytic
stripper, motoring and fired engine loads
Gilles et al. (2007) found that oils with larger fractions of high volatility materials
led to higher rates of oil consumption when examining consumption of numerous
different lubricating oils in a 2.2 liter DI Diesel engine. The findings were similar across
all loads and speeds tested. Oil consumption rates reported by Gilles et al. (2007) were
in the range of 30 to 45 g/hr for a DI Diesel engine running at 75 and 100% load, with
153
each load evaluated at 3000 and 4000 RPM. Normalizing the consumption rate to
engine displacement give rates of 14-21 g/hr liter. Consumption rates reported by
Yilmaz et al. (2004) were over a much wider operating range from no load to full load
at engine speeds ranging from 2500 to 5000 RPM. These authors found oil consumption
for a 2.0 liter, 4 cylinder engine to range from 7 to 82 g/hr. Again normalizing to
displacement gives rates of 4 to 41 g/hr liter. The highest rates were consistently found
at high loads and high speeds in both studies. Modeling vaporization rates alone, rather
than total oil consumption, Audette and Wong (1999) gave base vaporization rates on
the order of 1 g/hr per cylinder (.73 g/hr liter) for an engine operating at 2200 RPM and
full load with a similar bore, stroke, and compression ratio as the Isuzu 4HK1-TC used
in this HCCI work. Further analysis of reasonable mass transfer parameters used in the
study gave evaporation rates for the ±20 K liner temperature window ranging from .1 to
10 g/hr per cylinder (.073 to 7.3 g/hr liter) for the 1.38 liter displacement single cylinder
engine modeled. The range of evaporation rates spanning 2 orders of magnitude while
temperatures only span 40°C shows the sensitivity of evaporation rates to liner
temperatures.
Distillation curves of the oils used in the studies above were presented by all three
authors. The oil studied by Yilmaz et al. (2002) showed the most sensitivity to
temperature in the interval from 640 K to 740 K. In this temperature interval the percent
of total oil mass evaporated jumped from 10% to 80%. The ten different oil tested by
Gilles et al. (2007) showed very similar distillation profiles, with the bulk of the oil
evaporated between 625 K and 750 K. Audette and Wong (1999) presented a very
similar set of distillation curves for two different 15W-40 grade oils, the same grade of
oil as was used for the current HCCI work. The distillation curves presented by all
authors also show significant evaporation of mass, up to 10 %, below 600 K. Although
cylinder walls are at temperatures near 400 K - 450 K, the combustion gas temperatures
approach 1500 K. If a scenario of transient heat conduction is taking place in the
cylinder with each engine cycle, temperature distributions will fall between 400 K and
1500 K. Understanding this along with the distillation data support the likelihood that
154
oil vapor from both cylinder liner evaporation and mechanically generated droplet
evaporation within the hot cylinder gases occurs.
The total PM mass rate for ethanol testing ranged from .01 to 4.5 g/hr. The total
PM mass rate found during the pure hydrogen HCCI experiments ranged from .1 to .7
g/hr. With the lowest in-cylinder temperatures consistently corresponding to the lowest
exhaust PM mass rates. Motored PM mass rates are near .8 g/hr at 1500 RPM.
Comparing the mass rate of exhaust PM to the oil consumption rates found by Yilmaz
et al (2004) and Gilles et al. (2007) we see an inequality. A simple correlation between
oil consumption and PM total mass is not expected however, due to a number of points
that can be made to explain the difference. First, it should be pointed out that both
authors found a strong direct dependence of oil consumption on both engine load and
speed. The consumption rates reported by Yilmaz et al. (2004) were obtained at 75 and
100% rated load, and engine speeds of 3000 and 4000 RPM. The range of covered by
Gilles et al. (2007) extended from 2500 RPM to 5000 RPM over all engine loads. HCCI
work conducted here was done so at a low engine speed of 1500 RPM and low to
moderate engine loads. Second, blow-by will make significant contributions to oil
consumption in a high compression engine. The HCCI test engine had a crankcase
vented to ambient with a compression ratio of 18.5:1. The vented crankcase dumps
crankcase fumes and entrained oil to the ambient which adds to total oil consumption,
but would make no contribution to total particulate matter in the exhaust.
9.1.3
Conclusions
Particulate matter in the exhaust stream of a fully premixed ethanol HCCI engine
was characterized via TDMA analysis with thermal conditioning. Thermal conditioning
temperatures were swept from 40 to 110°C with maximum temperatures dictated by the
disappearance of measurable particles. Emissions from three fired loads were examined
along with a hot motoring condition. TDMA analysis has shown nucleation mode
particulate matter from this engine is composed of more than 98% volatile material.
These results are consistent across all load conditions. Particulate matter generated via
hot motoring of the engine has less volatile fraction, on the order of 85% at 110°C.
155
Combining this analysis with the results found in Sections 6.2.2, 7.2.2, 8.2.2, and
8.3.3, evidence suggests that particulate matter from this type of engine forms primarily
from lubricating oil and is highly dependant on cylinder temperatures and heat release
rates, thus independent of fuel. These results have shown changes in particulate
emissions can be induced via multiple SOC control strategies and emissions
consequences of each strategy consistently exhibit a dependence on peak cylinder
temperature and heat release rate.
Comparing the motored and loaded particle evaporation profiles and volatile
volume fractions, they author would like to put forth the hypothesis that a change in the
lubricating oil occurs during combustion. The lubricating oil in the engine is primarily
composed of hydrocarbons ranging in carbon number from C20 to C40 (Morgan, 2010).
As these hydrocarbons decompose into lighter components, vapor pressures of the
compounds will increase giving way to increased volatility. A non-sooting HCCI engine
creates exhaust conditions lacking carbonaceous soot on which to condense and adsorb
these organic vapors. They instead form nano-particles via homogeneous nucleation
during dilution and cooling of the exhaust.
9.2 FTIR Data
As advanced engine technology continually cuts regulated emissions, the exhaust
species once thought to be negligible begin to develop a significant contribution to the
whole of exhaust emissions. Recent studies by Reyes et al. (2006), Dukulis et al.
(2009), Kar and Cheng (2009), and Wallner and Frazee (2010) have all shown the
usefulness of FTIR techniques for engine emissions work. Reyes et al. (2006) cites
ammonia, formaldehyde, and nitrous oxide as species of particular interest.
Additionally, Wallner and Frazee (2010) highlight the rising concerns of oxygenated
biofuels leading to increases in aldehydes and alcohols in the exhaust stream. These
findings, combined with Kar and Cheng (2009) reporting the diminished response of
FID analyzers when used to examine highly oxygenated fuels, create a clear need for
more thorough examination of unregulated exhaust species. Although shortfalls in
practicality of FTIR as an engine exhaust analysis tool hindered widespread adoption at
156
its introduction in the late 1980’s, advances in the technology have lead to widespread
use and acceptance. According to Adachi (2000), the major obstacles FTIR faced when
compared to conventional NDIR, FID, and CLD instrumentation were; cross sensitivity,
response time, and accuracy of concentration. Adachi also notes that continued
development of the method has addressed many of these deficiencies, resulting in tools
and techniques that are now widely used across the automotive industry. Furthermore,
Reyes et al. (2006) notes several regulatory and standardization agencies that have
validated the technique for extractive gas sampling.
9.2.1
Experimental
Considering the data presented thus far, with particle size distributions composed
entirely of nucleation mode particles and understanding that these particles are formed
solely through gas to particle conversions, it is clear that a detailed examination of gas
phase exhaust components will aid in understanding PM formation and growth in HCCI
engines. FTIR data was collected during the TDMA experiments, with the engine
operating as previously explained. In addition to giving true “wet” concentration of
components in the exhaust stream, FTIR responds to a much wider variety of chemical
species than single component gas phase engine analyzers (CO, CO2, NOX, UHCs, O2).
Data on the chemical species listed in Table 18 were collected.
FTIR data was collected simultaneously with the TDMA data, Section 9.1 can be
referenced for engine operating conditions pertinent to this work. The FTIR instrument,
AVL model SESAM (System for Emission Sampling and Measurement), was operated
with reference spectra developed by the manufacturer specifically for ethanol fuel. The
spectrometer was operated with a sample flow of 10 lpm pulled through a heated filter
and sample line maintained at 185 °C. Before commencing data collection at each
operating condition, the engine was run for a minimum of fifteen minutes to ensure
stable operation. FTIR data was then collected at 1 hertz for the duration of the TDMA
experiments, roughly one and a half hours per engine load condition.
Comparisons are made to data obtained via conventional gas analyzers at the same
test conditions. Pertinent specifications of theses analyzers are shown in Table 17.
157
Table 17: Conventional gas analyzer descriptions
Analyzer Manufacturer Model
Method Range
9.2.2
CO
Horiba
VIA-510
NDIR
0-5000
ppm
CO2
Rosemont
880
NDIR
0-15%
NO/NO2
California
Analytical
Instruments
600-HCLD
CLD
0-10
ppm
HCs
J.U.M.
Engineering
3-300A
FID
0-1000
ppm
Results and Discussion
Along with the chemical species examined, Table 18 shows the coefficient of
variation for the data collected at each load condition. Chemical species with negative
concentrations were considered to at the noise level of the instrument and were omitted.
Most COVs were well below ten percent. The notable exceptions were nitric oxide,
nitrogen dioxide, and propene. The absolute magnitudes of these species are very low,
less than 1ppm. At these levels the signal to noise ratio of the instrument was quite high
resulting in a high degree of variation in the data. An acceptable degree of variation in
the data can be found at species concentrations above 1 ppm. Species with
concentrations below the sensitivity of the instrument are considered not measurable
and noted in Table 18.
158
Table 18: Chemical Species Examined via FTIR Spectroscopy
COVMidLoad 1
COVMidLoad 2
Chemical Species Abbreviation COVLowLoad
Water
H2O
0.2%
0.3%
0.2%
Carbon Dioxide
CO2
0.5%
0.2%
0.3%
Carbon Monoxide
CO
0.9%
1.0%
2.4%
Nitric Oxide
NO
11.8%
7.4%
4.1%
Nitrogen Dioxide
NO2
159.1%
141.0%
4.7%
Nitrous Oxide
N2O
3.3%
3.0%
2.4%
Ammonia
NH3
n/m
n/m
n/m
HCHO
0.9%
0.9%
1.9%
Methane
CH4
1.1%
1.1%
2.7%
Acetylene
C2H2
5.5%
5.0%
8.1%
Ethylene
C2H4
0.8%
0.9%
2.3%
Propene
C3H6
19.7%
98.0%
30.3%
Biethylene
C4H6
n/m
n/m
n/m
Acetaldehyde
MECHO
1.2%
0.7%
2.7%
Acrylaldehyde
C3H4O
n/m
n/m
n/m
Ethanol
ETOH
0.8%
0.4%
0.5%
Methanol
MEOH
4.1%
4.7%
6.9%
HCN
1.1%
1.1%
2.7%
Formaldehyde
Hydrogen Cyanide
Figure 84 shows FTIR spectroscopy data collected at three fired engine loads and a
motoring load. Results are presented on a ppm basis. To aid in comparison, emissions
data gathered via conventional gas bench analysis has also been converted to a ppm
basis. Motored data shows significant levels of H2O, CO, CO2, HCN in the air being
pumped through the engine. The water is a result of high humidity conditions in the lab
at the time of testing. Elevated levels of CO, CO2, and HCN found in the ambient
laboratory air result from inadequate lab ventilation and improperly sealed exhaust lines
on various test apparatus being used during or just prior to data collection. CO and CO2
are common exhaust gas species Additionally, Karlsson (2004) recently document
significant levels of HCN in engine exhaust.
159
Low Load
Mid Load 1
Mid Load 2
Motoring
100000
Concentration (ppm)
10000
1000
100
10
1
Ca
Ca rbo Wa
rb n D ter
on io
M xid
on e
N
N it oxi
itr ric de
og O
en x
N Di ide
itr ox
ou id
sO e
A xid
Fo mm e
rm o
al nia
de
h
M yde
et
A han
ce e
ty
l
Et ene
hy
l
Pr ene
Bi ope
A eth ne
ce yl
A tald ene
cr eh
yl
al yde
de
hy
d
E
th e
H
a
yd M n
ro e ol
ge th
n an
C y ol
an
id
e
0.1
Figure 84: Average emissions data collected via FTIR spectroscopy from ethanol
fueled HCCI combustion, 4 loads, 1500 RPM
For comparative purposes Table 19 summarizes data collected via conventional gas
bench analysis for NOX (CLD), CO (NDIR), CO2 (NDIR), HC (FID), and H2O
(calculated), at the same engine conditions. A 95% confidence interval was established
using a t-distribution. Data for calculating the confidence intervals were collected over
the course of 3 months with sample sizes of 10, 14, and 7 for the low, mid-1, and mid-2
loads respectively. Both analyses show data as ppm on a wet exhaust basis. H2O
exhaust fraction is based on the carbon balance relationship developed by Heywood
(1988). Because CO and CO2 were measured dry, a wet/dry correction has been applied
to them using the same technique.
A chemical balance assuming complete combustion was also used to calculate
expected CO2 and H2O values. Humid air was considered for this balance and the
results are shown in the columns of Table 19 labeled “Calc”. The calculated CO2 values
160
are very close to measured values, however calculated and measured H2O value exhibit
a larger discrepancy. This is a result of the bench measured H2O values actually coming
from a hydrogen balance calculation based on fuel flows and measured hydrocarbon
values which assumed dry intake air.
Table 19: Average ethanol HCCI exhaust gas composition as measured by
conventional gas analyzers
Load Condition
Low Load
Mid Load 1
Mid Load 2
Exhaust Gas
Bench
Calc.
Bench
Calc.
Bench
Calc.
Component
.5 ±0.1
1.2±0.1
7.1 ±1.7
NOX (ppm)
1970 ±70
1380 ±50
790 ±100
CO (ppm)
28650
CO2 (ppm)
31110 35840 ±140 37950 43740 ±980 44000
±440
810 ±50
1090 ±60
1320 ±60
HC (ppm,C2)
45040
66450
H2O (ppm)
79090
59940 55210 ±180 70110
±620
±1450
Table 20 shows the ratio of FTIR data divided by data collected with conventional
gas analyzers for each of the species shown. The significant variation in the NOX data is
due to the very low levels of NO, NO2, and N2O measured by the FTIR. With total
concentrations on a single ppm scale, measurements are approaching the noise level of
the instrument. Total hydrocarbons represent all compounds shown in Figure 84 from
formaldehyde to hydrogen cyanide.
161
Table 20: Ratio of ethanol HCCI exhaust gas measurements made via FTIR
compared with those collected through conventional gas analysis
Load Condition
Exhaust Gas
Low Load
Mid Load 1
Mid Load 2
Component (PPM)
FTIR
Ratio
FTIR
Ratio
FTIR
Ratio
1
166%
1
86%
3
43%
NOX
2120
108%
1550
113%
950
120%
CO
25340
88%
31090
87%
38460
88%
CO2
800
98%
1110
102%
1410
107%
HC (C2:EtOH)
1050
130%
1400
128%
1670
127%
THC (C2:HCs)
58260 129% 66860 121% 72360 109%
H2 O
Kar and Cheng (2009) have shown similar under reporting when comparing fast
response FID measurements to gas chromatograph (GC) data for an SI engine running
on ethanol. When blending oxygenates with gasoline they also found that the under
reporting of the FID shows a dependence on oxygenate concentration in the blend. For
the HCCI tests conducted here the hydrocarbons in the exhaust stream were 76%, 80%,
and 84% ethanol for the low, mid-1, and mid-2 load conditions. Comparing these
proportions with those reported by Kar and Cheng for pure ethanol SI operation at a
similar operating condition, 1500 RPM and 3.8 bar IMEP, we see almost twice as much
ethanol in the HCCI exhaust stream. The primary contributor to the high proportion of
ethanol species in the HCCI exhaust is the lesser ability of low temperature combustion
to oxidize all of the fuel in the cylinder. The likelihood that an HCCI engine has far
more unburned fuel than partially oxidized intermediates is higher than that of a hotter
burning SI engine. The second contributor is the more thorough measurements of the
GC, the authors list a calibration gas set consisting of 23 species of hydrocarbons. The
FTIR reference spectra used for HCCI analysis consisted of only 11 hydrocarbon
species.
Significant concentrations of formaldehyde and acetaldehyde, 40-60 ppm for each
compound, were also found at each operating condition. Schuetzle et al. (1994) have
identified methanol and ethanol as primary fuel precursors to formaldehyde and
acetaldehyde respectively. Wallner and Frazee (2010) report little dependence of
formaldehyde emissions on fuel oxygen content (alcohol proportions) for gasoline and
162
ethanol blends. However they do show a clear and direct dependence of acetaldehyde
emissions on oxygenated fuel proportion for both ethanol and butanol blends in a DI-SI
engine.
Wallner and Frazee also report much closer agreement between FTIR and CLD
measured NOX emissions, however this is at NOX levels nearly 2 orders of magnitude
higher than those found in our HCCI engine. The major NOX differences found in our
work are clearly an artifact of trying to measure ultra low NOX levels near the minimum
detection limit of the FTIR. In the same study Wallner and Frazee also presented NDIR
and FTIR obtained CO2 comparisons. They reported NDIR measurements on the order
or 5-10% below FTIR measurements at a variety of CO2 levels (engine loads) in the
range of 700 to 1200 gCO2/kW hr. The HCCI work conducted here does not agree with
those findings. In our work, FTIR measured CO2 was consistently 12% lower than
NDIR measurements. For this study CO2 levels were also close to those examined by
Wallner and Frazee. Comparing our NDIR and FTIR measured CO2 data with
calculated values of CO2 expected for complete combustion we see very reasonable
agreement between NDIR data and calculated values. Examining CO measurements,
our data shows FTIR response on the order of 10-20 % higher than NDIR methods. The
Wallner and Frazee study reported a difference of 5% lower. The cause for these
discrepancies is currently unknown.
9.2.3
Conclusions
FTIR spectroscopy techniques have emerged as a popular tool for investigating
regulated and non-regulated emissions in greater detail. The work presented here
applies these techniques to emissions from an HCCI engine running on ethanol fuel at 3
fired loads. Comparisons have also been made to conventional exhaust gas analyzers
used to examine CO, CO2, HCs, and NOX (NDIR, FID, and CLD).
The limitations of the FID for estimating total HC concentration are shown to be
relatively constant across all operating conditions with the FTIR reporting THCs nearly
30% higher than FID (C2) measurements. These hydrocarbons were primarily alcohols,
but significant fractions of aldehydes, methane, and ethylene were also present.
163
Increases in aldehyde formation in the emissions from alcohol fuels when compared
with non-oxygenated fuels are consistent with SI engine literature (Kar and Cheng,
2009).
When examining CO2, the FTIR based instrument tended to underreport the NDIR
based instrument results by about 12%. These results did not show good agreement with
other comparisons in the literature, however the body of literature is examining these
types of comparisons is very limited. Reasonable agreement was not found when
examining NOX, however this is likely due to the ultra low NOX levels encountered
with range of .5-5 ppm. Finally CO measurements were compared and gave results that
were within reason, however, also not in full agreement with the literature.
Current combustion emissions literature is in general agreement that the addition of
oxygenates to hydrocarbon fuels aids in cutting tailpipe HC emissions. To ensure that
the literature is correct in reporting significant decreases in unburned hydrocarbons,
when measured by FID techniques, it is important to develop an understanding of the
technique and possible limitations it may have.
164
Chapter 10
Summary and Conclusions
A 2005 model 4HK1-TC Isuzu, 5.2 liter, 4 cylinder Diesel engine was modified for
studies on HCCI combustion and emissions. A series of tests investigating three
common strategies for controlling SOC in HCCI engines were conducted. The primary
focus of the work was on the emissions consequences of these strategies. Brake specific
emissions of CO, HCs, NOX, and PM were presented along with detailed PM size and
mass distributions. Detailed analysis of the combustion process, gauged quantitatively
through SOC (CA10), burn duration (CA90 – CA10), peak in-cylinder temperature, and
peak heat release rate was also presented.
10.1 HCCI Combustion
The first of the experiments investigated the effect of intake temperature on
combustion phasing of ethanol fueled HCCI at 3 loads and documented emission trends.
Additionally, a similar set of experiments examining hydrogen fueled HCCI at 1 load
were executed. All testing was conducted at an engine speed of 1500 RPM. In
agreement numerous published works and modeling exercises, increases in intake
temperatures led to advanced SOC and shorter burn durations at all loads for both fuels.
Shorter burn durations and similar combustion efficiencies require higher rates of heat
release as was shown in the data. These higher rates of heat release gave increasing
peak in-cylinder temperatures in response to increased intake temperature across all test
cases.
The second set of tests examined changes in combustion parameters and emissions
trends in response to EGR rates. EGR rates were varied from 0 to 50 % of intake air by
volume. Again 3 ethanol fueled engine loads were tested at 1500 RPM. The findings of
the current literature were again validated, with increasing EGR rates leading to lower
rates of in-cylinder pressure rise. More retarded SOC was also encountered along with
longer combustion duration due to increased rates of EGR. Longer combustion
durations gave way to lower values of peak heat release. In-cylinder temperatures
remained relatively stable at low loads and increased with increasing EGR rates at
higher loads.
165
The third set of experiments conducted characterized combustion and emissions
when 2 fuels were used in varying proportions. Three loads were again examined at an
engine speed of 1500 RPM. The engine was operated in an HCCI mode with ethanol as
the primary fuel. Ethanol fueling was then removed, and the loss of power compensated
for by supplementation of hydrogen energy in amounts of 0, 5, 10, 15, 20, and 25 % of
engine output energy. Throughout all loads tested, as hydrogen energy proportion was
increased, SOC advanced. The advance was increasingly pronounced as load went up,
with the largest advance in SOC around 2.5 CAD. Compared to the other SOC control
strategies like intake temperature and EGR rate, supplemental hydrogen fueling
produced more modest changes in SOC timing. Along with more advanced combustion,
increased hydrogen fueling proportions led to shorter burn durations and higher peak
rates of heat release. Higher in-cylinder temperatures contributed to increasing
combustion efficiency as hydrogen fueling was increased.
10.2 HCCI Emissions
HCCI shows great promise as a low emissions combustion strategy for internal
combustion engines. The above research has systematically investigated some of the
most common SOC control strategies and documented their emissions consequences. In
an effort to clarify the root cause of changes in emissions behavior, emissions response
to the combustion parameters peak HRR, and peak temperature are shown. Data in
Chapter 6, Chapter 7, and Chapter 8 point to these as the primary drivers of emissions
behavior. Because different control strategies can be utilized to influence in-cylinder
combustion, the following summary plots illustrate the relationships between emissions
and control strategies. The relationships are independent of the control strategy itself.
The results show that emissions can be related to combustion parameters, although
these parameters are altered via diverse means. Presented below is a summary of the
emissions behavior of an ethanol and hydrogen fueled HCCI engine. The responses of
brake specific gas phase emissions to in-cylinder temperatures are shown in Figure 85,
Figure 86, and Figure 87. The data is from all of the testing conducted on both ethanol
and hydrogen fueled HCCI combustion.
166
The temperature dependence of the chemical kinetics which form these pollutants
is clearly shown in each figure. The beginnings of the exponentially increasing rate of
NO formation around 1800 K (Johansson, 2007) can be seen as in-cylinder
temperatures approach this level. Examining Figure 86 and Figure 87, we see the lower
limit for optimum combustion temperature highlighted. As in-cylinder temperatures
climb above 1500 K clear reductions in CO are seen. Further oxidation of hydrocarbons
to CO2 or hydrogen to water is also evident in Figure 87 as temperatures increase. These
trends are also manifested in increasing combustion efficiency. In all cases examined,
the combustion efficiency, calculated from unburned fuel, peaked at the highest incylinder temperatures. Combustion efficiencies ranged from 70 to 95% for ethanol
HCCI and from 94 to 96% for hydrogen HCCI. The lowest combustion efficiencies
represented cases of intermittent misfire.
0.2
0.18
BSNOX (g/kW hr)
0.16
0.14
Pure H2
EGR
Intake Temp
EtOH:H2
0.12
0.1
0.08
0.06
0.04
0.02
0
1200
1300
1400
1500
1600
1700
1800
Peak Cylinder Temperature (K)
Figure 85: BSNOX v. peak cylinder temperatures for ethanol and hydrogen HCCI
with SOC controlled by multiple strategies
167
120
EGR
Intake Temp
EtOH:H2
BSCO (g/kW hr)
100
80
60
40
20
0
1200
1300
1400
1500
1600
1700
1800
Peak Cylinder Temperature (K)
Figure 86:BSCO v. peak cylinder temperatures for ethanol and hydrogen HCCI
with SOC controlled by multiple strategies
120
BSHC or H2 (g/kW hr)
100
80
EGR
Intake Temp
EtOH:H2
Pure H2
60
40
20
0
1200
1300
1400
1500
1600
1700
1800
Peak Cylinder Temperature (K)
Figure 87: BSHC of BSH2 v. peak cylinder temperatures for ethanol and hydrogen
HCCI with SOC controlled by multiple strategies
168
Inspecting all BSPM data, the clearest relationship was shown between BSPM and
peak HRR. It is shown in Figure 88 with the highly variable intake temperature data,
especially at low loads, highlighted.
0.30
EGR
H2
EtOH:H2
Intake Temp
BSPM (g/kW hr)
0.25
0.20
0.15
0.10
0.05
0.00
0
50
100
150
200
Peak Heat Release Rate (J/CAD)
Figure 88: BSPM v. peak HRR for ethanol and hydrogen HCCI with SOC
controlled by multiple strategies
A multiple regression analysis examining the relationship between total particulate
mass and the combustion parameters; SOC, combustion duration, peak temperature, and
peak heat release rate was conducted. Utilizing ethanol HCCI data from the variable
intake temperature, variable EGR, and hydrogen supplemented HCCI tests, an initial R2
of .51 was found with a sample size of 44. As in the earlier regression analysis
examining only the variable intake temperature data, this analysis was repeated with the
omission of the highly variable low load condition. The new regression gave a R2 of .66
with sample size of 38. Taking into account the sporadic behavior of the variable intake
temperature data explained in section 6.3, a final regression was conducted omitting all
of this data. The final R2 jumped sharply to .82. Omitting all of the variable intake
temperature data still left a sample size of 30, due to the contributions of the EGR and
supplemental hydrogen data.
169
Table 21 is a correlation matrix showing the relationship of emissions to common
combustion parameters. The trends were established though the analysis of ethanol
HCCI, ethanol HCCI with supplemental hydrogen fueling, and pure hydrogen HCCI. In
the BSFC column an optimization point was found in most cases, where increases in the
independent variable lead to decreases in BSFC initially, however a minima was
eventually reached and BSFC started to increase with increasing SOC. The arrows
indicate the response of the emissions characteristics to increases in peak temperature,
peak heat release rate, and combustion duration respectively. In the case of SOC the
arrows indicate the response of the emissions characteristics to delays in SOC.
Table 21: Correlation matrix relating emissions to combustion properties in fully
premixed HCCI combustion of ethanol and hydrogen
PM Total
BSPM
BSFC
BSHC
BSCO
BSNOX
Mass
(g/kW
hr)
(g/kW
hr)
(g/kW
hr)
(g/kW
hr)
(g/kW
hr)
(µg/m3)
Peak
Temp (K)
Peak
HRR
(J/CAD)
SOC,
CA 10
(ºATDC)
Duration
(CAD)
↑
↑
↓
↓
↑
↑
↓
↓
↕ ↓ ↓
↕ ↓ ↓
↕ ↑* ↑*
↕ ↑* ↑*
↑
↑
↓
↓
*Trends are not clearly defined when EGR is employed to control combustion
phasing
Analysis of particle volatility was conducted through TDMA techniques to gain
further understanding of the composition, and ultimately sources, of fully premixed
ethanol HCCI particulate matter. The results suggested lubricating oil as the primary
contributor to particulate matter in these types of engines.
170
Analysis of the lubricating oil consumption literature points to oil vaporization
from the cylinder walls, reverse gas flow and piston throw-off as the primary routes of
oil consumption at the load conditions found throughout our study. A relative
insensitivity of the oil film to combustion gas temperatures, in terms of cycle by cycle
resolution, in CI and SI engines has been published in the literature is likely true for this
HCCI work as well. The cylinder liner temperature will however increase as increases
in in-cylinder temperatures require more heat flow to the coolant circuit. Furthermore
increased bulk cylinder gas temperatures enhance evaporation of oil droplets generated
via reverse gas flow and piston throw-off.
Total particulate mass rates found here were somewhat less than published oil
consumption rates. A direct match in these rates was not expected however. Because the
higher volatility fractions of the oil are evaporating at the highest rate, the likelihood
that these species achieve full or partial oxidation in a low temperature combustion
environment is reasonable. This hypothesis combined with the substantial blow-by of
the high compression test engine, which routes crankcase fumes out to the ambient,
serves to compensate for the difference between published oil consumption rates and
the total particulate mass rates found here.
In summary, these findings strongly suggest that although essentially free of
accumulation mode (soot) particles, nucleation mode particulate matter is present in
significant mass and number in the exhaust of fully premixed HCCI engines. These
results are also indicative of the behavior to be expected from other modern low sooting
engines. Precursors to this particulate matter are primarily derived from more volatile
species in the lubricating oil. The abundance of volatile precursor and lack of adsorption
and condensation sites create ideal conditions for homogeneous nucleation.
171
Bibliography
Aakko, P., & Nylund, N. Particle emissions at moderate and cold temperatures using
different fuels.
Abd-Alla, G. H. (2002). Using exhaust gas recirculation in internal combustion engines:
A review. Energy Conversion and Management, 43(8), 1027-1042.
Abdul-Khalek, I., Kittelson, D., & Brear, F. (2000). Nanoparticle growth during dilution
and cooling of diesel exhaust: Experimental investigation and theoretical
assessment. SAE Technical Paper Series, Paper No. 2000-01-0515
Abdul-Khalek, I., Kittelson, D. B., & Brear, F. The influence of dilution conditions on
diesel exhaust particle size distribution measurement. International Congress and
Exposition, Detroit, MI. , Paper No. 1999-01-1142
Adachi, M. (2000). Emission measurement techniques for advanced powertrains.
Measurement Science and Technology, 11(10), R113-R129.
Agarwal, J. K., & Sem, G. J. (1980). Continuous flow, single-particle-counting
condensation nucleus counter. Journal of Aerosol Science, 11(4), 343-357.
Al-Hasan, M. (2003). Effect of ethanol-unleaded gasoline blends on engine
performance and exhaust emission. Energy Conversion and Management, 44(9),
1547-1561.
Apple, J., Gladis, D., Watts, W., & Kittelson, D. (2009). Measuring diesel ash
emissions and estimating lube oil consumption using a high temperature oxidation
method. SAE Technical Paper Series, (Paper No. 2009-01-1843)
Audette, W. E., & Wong, V. W. (1999). A model for estimating oil vaporization from
the cylinder liner as a contributing mechanism to engine oil consumption. SAE
Technical Paper Series, Paper No. 1999-01-1520
Ayar, G. (2006). Compressed natural gas (CNG) in fueled systems and the significance
of CNG in vehicular transportation. Energy Sources Part A: Recovery, Utilization
& Environmental Effects, 28(7), 639-641.
Biswas, S., Fine, P. M., Geller, M. D., Hering, S. V., & Sioutas, C. (2005). Performance
evaluation of a recently developed water-based condensation particle counter.
Aerosol Science and Technology, 39(5), 419-427.
Burtscher, H. (2005). Physical characterization of particulate emissions from diesel
engines: A review. Journal of Aerosol Science, 36(7), 896-932.
172
Burtscher, H., Künzel, S., & Hüglin, C. (1998). Characterization of particles in
combustion engine exhaust. Journal of Aerosol Science, 29(4), 389-396.
Caton, P. A., & Pruitt, J. T. (2009). Homogeneous charge compression ignition of
hydrogen in a single-cylinder diesel engine. International Journal of Engine
Research, 10, 45-63.
Chen, D. R., Pui, D. Y. H., Hummes, D., Fissan, H., Quant, F. R., & Sem, G. J. (1998).
Design and evaluation of a nanometer aerosol differential mobility analyzer (nanoDMA). Journal of Aerosol Science, 29(5-6), 497-509.
Cho, H. M., & He, B. (2007). Spark ignition natural gas engines—A review. Energy
Conversion and Management, 48(2), 608-618.
Combustion modeling in reciprocating engines(1980). In General Motors Corporation.
Research Laboratories (Ed.),1. New York: New York : Plenum Press.
Conaire, M. O., Curran, H. J., Simmie, J. M., Pitz, W. J., & Westbrook, C. K. (2004). A
comprehensive modeling study of hydrogen oxidation. International Journal of
Chemical Kinetics, 36, 603-622.
Dec, J. E. (1997). A conceptual model of DI diesel combustion based on laser-sheet
imaging. SAE Technical Paper Series, (Paper No. 970873)
Dec, J., & Sjöberg, M. (2003). A parametric study of HCCI combustion-the sources of
emissions at low loads and the effects of GDI fuel injection. SAE Technical Paper
Series, Paper No. 2003-01-0752
Dec, J., Sjöberg, M., & Hwang, W. (2009). Isolating the effects of EGR on HCCI heatrelease rates and NOX emissions. SAE International Journal of Engines, 2(2), 5870.
Deluga, G. A., Salge, J. R., Schmidt, L. D., & Verykios, X. E. (2004). Renewable
hydrogen from ethanol by autothermal reforming. Science, 303(5660), 993-997.
Dukulis, I., Pirs, V., Jesko, Z., Birkavs, A., & Birzietis, G. (2009). Testing of
automobile VW golf operating on three different fuels. 8th International Scientific
Conference Engineering for Rural Development, Jelgava (Latvia). 7-13.
Epping, K., Aceves, S. M., Bechtold, R., & Dec, J. (2002). The potential of HCCI
combustion for high efficiency and low emissions. SAE Technical Paper Series,
Paper No. 2002-01-1923
Flowers, D. L., Martinez-Frias, J., Espinosa-Loza, F., Dibble, R., Kristic, M., Bining,
A., et al. (2005). Development and testing of a 6-cylinder HCCI engine for
173
distributed generation. Internal Combustion Engine Division of ASME 2005 Fall
Technical Conference, Ottawa, Canada.
Franklin, L. M., Bika, A. S., Watts, W. F., & Kittelson, D. B. (2010). Comparison of
water and butanol based CPCs for examining diesel combustion aerosols. Aerosol
Science and Technology, 44(8), 629.
Gilles, A., Pidol, L., Delvigne, T., Courtois, O., & Deconninck, B. (2007). Influence of
lubricant formulations on engine oil consumption using radionuclide technique.
SAE Technical Paper Series, Paper No. 2007-01-1981
Gomes Antunes, J. M., Mikalsen, R., & Roskilly, A. P. (2008). An investigation of
hydrogen-fuelled HCCI engine performance and operation. International Journal
of Hydrogen Energy, 33(20), 5823-5828.
Gomes Antunes, J. M., Mikalsen, R., & Roskilly, A. P. (2009). An experimental study
of a direct injection compression ignition hydrogen engine. International Journal
of Hydrogen Energy, 34(15), 6516-6522.
Griffiths, J. F., Halford-Maw, P. A., & Rose, D. J. (1993). Fundamental features of
hydrocarbon autoignition in a rapid compression machine. Combustion and Flame,
95(3), 291-306.
Handbook of air pollution from internal combustion engines : Pollutant formation and
control(1998). In Sher E. (Ed.), . Boston: Boston : Academic Press.
Hasegawa, R., & Yanagihara, H. (2003). HCCI combustion in diesel engine. SAE
Technical Paper Series, Paper No. 2003-01-0745
He, B., Jian-Xin Wang, Hao, J., Yan, X., & Xiao, J. (2003). A study on emission
characteristics of an EFI engine with ethanol blended gasoline fuels. Atmospheric
Environment, 37(7), 949-957.
Hering, S. V., & Stolzenburg, M. R. (2005). A method for particle size amplification by
water condensation in a laminar, thermally diffusive flow. Aerosol Science and
Technology, 39(5), 428-436.
Hering, S. V., Stolzenburg, M. R., Quant, F. R., O'Berreit, D. R., & Keady, P. B.
(2005). A laminar-flow, water-based condensation particle counter (WCPC).
Aerosol Science and Technology, 39(7), 659-672.
Hermann, M., Wehner, B., Bischof, O., Han, H. -., Krinke, T., Liu, W., et al. (2007).
Particle counting efficiencies of new TSI condensation particle counters. Journal of
Aerosol Science, 38(6), 674-682.
174
Heywood, J. B. (1988). Internal combustion engine fundamentals. New York: New
York : McGraw-Hill.
Hilliard, J. C., & Wheeler, R. W. (1979). Nitrogen dioxide in engine exhaust. SAE
Transactions, 88(Paper No. 790691)
Hosseini, V., & Checkel, M. D. (2006). Using reformer gas to enhance HCCI
combustion of CNG in CFR engine. SAE Technical Paper Series, Paper No. 200601-3247
Hosseini, V., & Checkel, M. D. (2007). Effect of reformer gas on HCCI combustion part I: High octane fuels. SAE Technical Paper Series, Paper No. 2007-01-0208
Hosseini, V., & Checkel, M. D. (2007). Effect of reformer gas on HCCI combustion part II: Low octane fuels. SAE Technical Paper Series, Paper No. 2007-01-0206
Hosseini, V., & Checkel, M. D. (2008). Reformer gas composition effect on HCCI
combustion of n-heptane, iso-octane, and natural gas. SAE Technical Paper Series,
Paper No. 2008-01-0049
Hsieh, W. D. (2002). Engine performance and pollutant emission of an SI engine using
ethanol-gasoline blended fuels. Atmospheric Environment, 36(3), 403.
Hyvönen, J., Haroldsson, G., & Johansson, B. (2004). Balancing cylinder-to-cylinder
variations in a multi-cylinder VCR-HCCI engine. SAE Technical Paper Series,
Paper No. 2004-01-1897
Iida, K., Stolzenburg, M. R., McMurry, P. H., Smith, J. N., Quant, F. R., Oberreit, D.
R., et al. (2008). An ultrafine, water-based condensation particle counter and its
evaluation under field conditions. Aerosol Science and Technology, 42(10), 862871.
Ishibashi, Y. (2000). Basic understanding of activated radical combustion and it twostroke engine application and benefits. SAE Technical Paper Series, Paper No.
2000-01-1836
Johansson, B. (2007). Homogeneous charge compression ignition: The future of IC
engines. International Journal of Vehicle Design, 44(1/2), 1-19.
Johnson, T., Caldow, R., Pocher, A., Mirme, A., & Kittelson, D. (2004). A new
electrical mobility particle sizer spectrometer for engine exhaust particle
measurements. SAE Technical Paper Series, Paper No. 2001-01-1341
175
Johnson, J. P., Kittelson, D. B., & Watts, W. F. (2005). Source apportionment of diesel
and spark ignition exhaust aerosol using on-road data from the minneapolis
metropolitan area. Atmospheric Environment, 39(11), 2111-2121.
Jung, H., Kittelson, D. B., & Zachariah, M. R. (2006). Characteristics of SME
biodiesel-fueled diesel particle emissions and the kinetics of oxidation.
Environmental Science & Technology, 40(16), 4949-4955.
Juttu, S., Thipse, S. S., Marathe, N. V., & Gajendra Babu, M. K. (2007). Homogeneous
charge compression ignition (HCCI): A new concept for near zero NOX and
particulate matter (PM) from diesel engine combustion. SAE Technical Paper
Series, Paper No. 2007-26-020
Kaiser, E. W., Yang, J., Culp, T., & Maricq, M. M. (2002). Homogeneous charge
compression ignition engine-out emissions—does flame propagation occur in
homogeneous charge compression ignition? International Journal of Engine
Research, 3(4), 185-195.
Kaminski, D. A. (2005). In Jensen M. K. (Ed.), Introduction to thermal and fluid
engineering. Hoboken, N.J.: Hoboken, N.J. : Wiley.
Kar, K., & Cheng, W. K. (2009). Speciated engine-out organic gas emissions from a
PFI-SI engine operating on ethanol/gasoline mixtures. SAE International Journal
of Fuels and Lubricants, 2(2), 91-101.
Karlsson, H. L. (2004). Ammonia, nitrous oxide and hydrogen cyanide emissions from
five passenger vehicles. Science of the Total Environment, 334-335, 125-132.
doi:DOI: 10.1016/j.scitotenv.2004.04.061
Kasper, M. (2005). Sampling and measurement of nanoparticle emissions for type
approval and field control. SAE Technical Paper Series, Paper No. 2005-26-013
Kayes, D., & Hochgreb, S. (1999). Mechanisms of particulate matter formation in
spark-ignition engines. 1. effect of engine operating conditions. Environmental
Science & Technology, 33(22), 3957-3967.
Kayes, D., & Hochgreb, S. (1999). Mechanisms of particulate matter formation in
spark-ignition engines. 2. effect of fuel, oil, and catalyst parameters. Environmental
Science & Technology, 33(22), 3968-3977.
Kim, D., Gautam, M., & Gera, D. (2002). Parametric studies on the formation of diesel
particulate matter via nucleation and coagulation modes. Journal of Aerosol
Science, 33(12), 1609-1621.
176
Kimura, S., Aoki, O., Ogawa, H., & Muranaka, Y. (1999). New combustion concept for
ultra-clean and high-efficiency small DI diesel engines. SAE Technical Paper
Series, Paper No. 1999-01-3681
Kitamura, T., Senda, J., & Fujimoto, H. (2002). Mechanism of smokeless diesel
combustion with oxygenated fuels based on the dependence of the equivalence
ratio and temperature on soot particle formation. International Journal of Engine
Research, 3(4), 223-248.
Kittelson, D., Watts, W., & Johnson, J. (2002). DIESEL AEROSOL SAMPLING
METHODOLOGY - CRC E-43, final report No. E-43)
Kittelson, D. B. (1998). Engines and nanoparticles: A review. Journal of Aerosol
Science, 29(5-6), 575-588.
Knothe, G., Sharp, C. A., & Ryan, T. W. (2006). Exhaust emissions of biodiesel,
petrodiesel, neat methyl esters, and alkanes in a new technology engine†. Energy &
Fuels, 20(1), 403-408.
Knutson, E. O., & Whitby, K. T. (1975). Aerosol classification by electric mobility:
Apparatus, theory, and applications. Journal of Aerosol Science, 6(6), 443-451.
Kuehn, T. H. (. (1998). In Ramsey J. W. (. W. )., Threlkeld J. L. (Eds.), Thermal
environmental engineering. Upper Saddle River, N.J.: Upper Saddle River, N.J. :
Prentice Hall.
Ladommatos, N., Abdelhalim, S., & Zhao, H. (2000). The effects of exhaust gas
recirculation on diesel combustion and emissions. International Journal of Engine
Research, 1(1), 107-126.
Lee, D., Miller, A., Kittelson, D., & Zachariah, M. R. (2006). Characterization of metalbearing diesel nanoparticles using single-particle mass spectrometry. Journal of
Aerosol Science, 37(1), 88-110.
Lemmon, E. W., & Goodwin, A. R. H. (2000).
Critical properties and vapor pressure equation for alkanes C[sub n]H[sub 2n + 2]:
Normal alkanes with n <= 36 and isomers for n = 4 through n = 9. Journal of
Physical and Chemical Reference Data, 29(1), 1-39.
Liu, B. Y. H., Pui, D. Y. H., Whitby, K. T., Kittelson, D. B., Kousaka, Y., & McKenzie,
R. L. (1978). The aerosol mobility chromatograph: A new detector for sulfuric acid
aerosols. Atmospheric Environment (1967), 12(1-3), 99-104.
Liu, B. Y. H., & Kim, C. S. (1977). On the counting efficiency of condensation nuclei
counters. Atmospheric Environment (1967), 11(11), 1097-1100.
177
Liu, B. Y. H., & Pui, D. Y. H. (1974). A submicron aerosol standard and the primary,
absolute calibration of the condensation nuclei counter. Journal of Colloid and
Interface Science, 47(1), 155-171.
Liu, W., Kaufman, S. L., Osmondson, B. L., Sem, G. J., Quant, F. R., & Oberreit, D. R.
(2006). Water-based condensation particle counters for environmental monitoring
of ultrafine particles. Journal of the Air and Waste Management Association, 56(4),
444-455.
Lü, X., Chen, W., & Huang, Z. (2005). A fundamental study on the control of the HCCI
combustion and emissions by fuel design concept combined with controllable
EGR. part 1. the basic characteristics of HCCI combustion. Fuel, 84(9), 1074-1083.
Lü, X., Chen, W., & Huang, Z. (2005). A fundamental study on the control of the HCCI
combustion and emissions by fuel design concept combined with controllable
EGR. part 2. effect of operating conditions and EGR on HCCI combustion. Fuel,
84(9), 1084-1092.
Mack, J. H., Aceves, S. M., & Dibble, R. W. (2009). Demonstrating direct use of wet
ethanol in a homogeneous charge compression ignition (HCCI) engine. Energy,
34(6), 782-787.
Marinov, N. M. (1999). A detailed chemical kinetic model for high temperature ethanol
oxidation. International Journal of Chemical Kinetics, 31, 183-220.
Martinez-Frias, J., Aceves, S. M., & Flowers, D. L. (2007). Improving ethanol life cycle
energy efficiency by direct utilization of wet ethanol in HCCI engines. Journal of
Energy Resources Technology, 129(4), 332-337.
Martinez-Frias, J., Aceves, S. M., Flowers, D. L., Smith, J. R., & Dibble, R. (2002).
Thermal charge conditioning for optimal HCCI engine operation. Journal of
Energy Resources Technology, 124(1), 67-75.
Mathis, U., Ristimäki, J., Mohr, M., Keskinen, J., Ntziachristos, L., Samaras, Z., et al.
(2004). Sampling conditions for the measurement of nucleation mode particles in
the exhaust of a diesel vehicle. Aerosol Science & Technology, 38(12), 1149-1160.
McClave, J. T., & Sincich, T. (2003). Statistics (9th ed.). Upper Saddle River, N.J.:
Prentice Hall.
McMurry, P. H. (2000). The history of condensation nucleus counters. Aerosol Science
& Technology, 33(4), 297-322.
178
Miller, A. L., Stipe, C. B., Habjan, M. C., & Ahlstrand, G. G. (2007). Role of
lubrication oil in particulate emissions from a hydrogen-powered internal
combustion engine. Environmental Science and Technology, 41(19), 6828-6835.
Milovanovic, N., Rui, C., & Turner, J. (2004). Influence of the variable valve timing
strategy on the control of a homogeneous charge compression (HCCI) engine. SAE
Technical Paper Series, Paper No. 2004-01-1899
Misztal, J., Xu, H. M., Wyszynski, M. L., Price, P., Stone, R., & Qiao, J. (2009). Effect
of injection timing on gasoline homogeneous charge compression ignition
particulate emissions. International Journal of Engine Research, 10(6), 419-430.
Misztal, J., Xu, H., Tsolakis, A., Wyszynski, M. L., Constandinides, G., Price, P., et al.
(2009). Influence of inlet air temperature on gasoline HCCI particulate emissions.
Combustion Science and Technology, 181(5), 695-709.
Moran, M. J., & Shapiro, H. N. (2000). In Shapiro H. N. (Ed.), Fundamentals of
engineering thermodynamics (4th ed. ed.). Hoboken, N.J.: Hoboken, N.J. : Wiley.
Mordas, G., Manninen, H. E., Petäjä, T., Aalto, P. P., Hämeri, K., & Kulmala, M.
(2008). On operation of the ultra-fine water-based CPC TSI 3786 and comparison
with other TSI models (TSI 3776, TSI 3772, TSI 3025, TSI 3010, TSI 3007).
Aerosol Science and Technology, 42(2), 152-158.
Morgan, N. (2010). Composition of shell rotella 15W-40 lubricating oil
Müller, M. (2010). General air fuel ratio and EGR definitions and their calculation from
emissions. SAE Technical Paper Series,
Najt, P. M., & Foster, D. E. (1983). Compression-ignited homogeneous charge
combustion. SAE Technical Paper Series, Paper No. 830264
Nakano, M., Mandokoro, Y., Kubo, S., & Yamazaki, S. (2000). Effects of exhaust gas
recirculation in homogeneous charge compression ignition engines. International
Journal of Engine Research, 1(3), 269-279.
Ng, C., K.W., & Thomson, M. J. (2007). Modelling of the effect of fuel reforming and
EGR on the acceptable operating range of an ethanol HCCI engine. International
Journal of Vehicle Design, 44(1-2), 107-123.
Ni, M., Leung, D. Y. C., & Leung, M. K. H. (2007). A review on reforming bio-ethanol
for hydrogen production. International Journal of Hydrogen Energy, 32(15), 32383247.
179
Noguchi, M., Tanaka, Y., Tanaka, T., & Takeuchi, Y. (1979). A study on gasoline
engine combustion by observation of intermediate reactive products during
combustion. SAE Technical Paper Series, Paper No. 790840
Onishi, S., Han Jo, S., Shoda, K., Do Jo, P., & Kato, S. (1979). Active thermoatmosphere combustion (ATAC): A new combustion process for internal
combustion engines. SAE Technical Paper Series, Paper No. 790501
Orsini, D. A., Wiedensohler, A., Stratmann, F., & Covert, D. S. (1999). A new volatility
tandem differential mobility analyzer to measure the volatile sulfuric acid aerosol
fraction. Journal of Atmospheric and Oceanic Technology, 16(6), 760-772.
Park, K., Cao, F., Kittelson, D. B., & McMurry, P. H. (2003). Relationship between
particle mass and mobility for diesel exhaust particles. Environmental Science and
Technology, 37(3), 577-583.
Peineke, C., Attoui, M. B., & Schmidt-Ott, A. (2006). Using a glowing wire generator
for production of charged, uniformly sized nanoparticles at high concentrations.
Journal of Aerosol Science, 37(12), 1651-1661.
Peng, Y., Tan, M., Guo, L., Liu, F., Li, H., & Guo, Y. (2007). Study the ethanol
SI/HCCI combustion mode transition by using the fast thermal management
system. Chinese Science Bulletin, 52(19), 2731-2736.
Petäjä, T., Mordas, G., Manninen, H. E., Aalto, P. P., Hämeri, K., & Kulmala, M.
(2006). Detection efficiency of a water-based TSI condensation particle counter
3785. Aerosol Science & Technology, 40, 1090-1097.
Peter H. McMurry, Hiroshi Takano, Gary R. Anderson. (1983). Study of the ammonia
(gas)-sulfuric acid (aerosol) reaction rate. Environmental Science & Technology,
17(6), 347-352.
Peucheret, S., Wyszyński, M. L., Lehrle, R. S., Golunski, S., & Xu, H. (2005). Use of
catalytic reforming to aid natural gas HCCI combustion in engines: Experimental
and modelling results of open-loop fuel reforming. International Journal of
Hydrogen Energy, 30(15), 1583-1594.
Poulopoulos, S. G. (2001). Regulated and unregulated emissions from an internal
combustion engine operating on ethanol-containing fuels. Atmospheric
Environment, 35(26), 4399.
Poulopoulos, S. G. (2003). The effect of adding oxygenated compounds to gasoline on
automotive exhaust emissions. Journal of Engineering for Gas Turbines and
Power, 125(1), 344.
180
Price, P., Stone, R., Misztal, J., Xu, H., Wyszynski, M., Wilson, T., et al. (2007).
Particulate emissions from a gasoline homogeneous charge compression ignition
engine. SAE Technical Paper Series, Paper No. 2007-01-0209
Quant, F. R., Flagan, R. C., & Horton, K. D. (1993). Implementation of a scanning
mobility particle sizer (SMPS). Journal of Aerosol Science, 24(Supplement 1),
S83-S84.
Rader, D. J., & McMurry, P. H. (1986). Application of the tandem differential mobility
analyzer to studies of droplet growth or evaporation. Journal of Aerosol Science,
17(5), 771-787.
Rahbari, A. (2008). The effect of EGR on HCCI engines using ethanol as fuel. SAE
Technical Paper Series, Paper No. 2008-01-2409
Reineking, A., & Porstendörfer, J. (1986). Measurements of particle loss functions in a
differential mobility analyzer (TSI, model 3071) for different flow rates. Aerosol
Science & Technology, 5, 483-485.
Reyes, F., Grutter, M., & Jazcilevich, A. G., R. (2006). Technical note: Analysis of nonregulated vehicular emissions by extractive FTIR spectrometry: Tests on a hybrid
car in mexico city. Atmospheric Chemistry and Physics, 6(12), 5339-5346.
Ristimäki, J., Vaaraslahti, K., Lappi, M., & Keskinen, J. (2007). Hydrocarbon
condensation in heavy-duty diesel exhaust. Environmental Science Technology,
41(18), 6397.
Ristovski, Z. D., Morawska, L., Hitchins, J., Thomas, S., Greenaway, C., & Gilbert, D.
(2000). PARTICLE EMISSIONS FROM COMPRESSED NATURAL GAS
ENGINES. Journal of Aerosol Science, 31(4), 403-413.
Rönkkö, T., Virtanen, A., Vaaraslahti, K., Keskinen, J., Pirjola, L., & Lappi, M. (2006).
Effect of dilution conditions and driving parameters on nucleation mode particles
in diesel exhaust: Laboratory and on-road study. Atmospheric Environment, 40(16),
2893-2901.
Rosati, M. F., & Aleiferis, P. G. (2009). Hydrogen SI and HCCI combustion in a directinjection optical engine. SAE International Journal of Engines, 2(1), 1710-1736.
Sakurai, H., Tobias, H. J., Park, K., Zarling, D., Docherty, K. S., Kittelson, D. B., et al.
(2003). On-line measurements of diesel nanoparticle composition and volatility.
Atmospheric Environment, 37(9-10), 1199-1210.
Schmidt-Ott, A., Schurtenberger, P., & Siegmann, H. C. (1980). Enormous yield of
photoelectrons from small particles. Physical Review Letters, 45(15), 1284-1287.
181
Schneider, J. (2005). Nucleation particles in diesel exhaust: Composition inferred from
in situ mass spectrometric analysis. Environmental Science Technology, 39(16),
6153.
Schreiber, D., Forss, A. M., Mohr, M., & Dimopoulos, P. (2007). Particle
characterization of modern CNG, gasoline, and diesel passenger cars. SAE
Technical Paper Series, Paper No. 2007-24-0123
Schuetzle, D., Siegl, W. O., Jensen, T. E., Dearth, M. A., Kaiser, E. W., Gorse, R., et al.
(1994). The relationship between gasoline composition and vehicle hydrocarbon
emissions: A review of current studies and future research needs. Environmental
Health Perspectives, 102(, Supplement 4: Risk Assessment of Urban Air:
Emissions, Exposure, Risk Identification, and Risk Quantitation), 3-12.
Shibata, G., & Urushihara, T. (2007). Auto-ignition characteristics of hydrocarbons and
development of an HCCI fuel index. SAE Technical Paper Series, Paper No. 200701-0220
Shin, K., Tateishi, Y., & Furuhama, S. (1983). Measurement of oil film thickness
between piston ring and cylinder. SAE Transactions, 92(Paper No. 830068)
Shudo, T., Ono, Y., & Takahashi, T. (2003). Ignition control by DME-reformed gas in
HCCI combustion of DME. SAE Technical Paper Series, Paper No. 2003-01-1824
Shudo, T., & Takahashi, T. (2004). Influence of reformed gas composition on HCCI
combustion of onboard methanol-reformed gases. SAE Technical Paper Series,
Paper No. 2004-01-1908
Shudo, T., & Yamada, H. (2007). Hydrogen as an ignition-controlling agent for HCCI
combustion engine by suppressing the low-temperature oxidation. International
Journal of Hydrogen Energy, 32(14), 3066-3072.
Sjöberg, M., Dec, J. E., & Hwang, W. (2007). Thermodynamic and chemical effects of
EGR and it constituents on HCCI autoignition. SAE Technical Paper Series, Paper
No. 2007-01-0207
Stanglmaier, R. H., & Roberts, C. E. (1999). Homogeneous charge compression
ignition (HCCI): Benefits, compromises, and future applications. SAE Technical
Paper Series, Paper No. 1999-01-3682
Stone, R., 1955-. (1993). Introduction to internal combustion engines. Warrendale, PA:
Warrendale, PA : Society of Automotive Engineers.
Structured catalysts and reactors(1998). In Cybulski A., Moulijn J. A. (Eds.), . New
York: New York : M. Dekker.
182
Surawski, N. C., Miljevic, B., Roberts, B. A., Modini, R. L., Situ, R., Brown, R. J., et al.
(2010). Particle emissions, volatility, and toxicity from an ethanol fumigated
compression ignition engine. Environmental Science & Technology, 44(1), 229235.
Swami Nathan, S., Mallikarjuna, J. M., & Ramesh, A. (2010). Effects of charge
temperature and exhaust gas re-circulation on combustion and emission
characteristics of an acetylene fuelled HCCI engine. Fuel, 89(2), 515-521.
Tanner, R. L. (1988). Atmospheric chemistry of aldehydes: Enhanced peroxyacetyl
nitrate formation from ethanol-fueled vehicular emissions. Environmental Science
Technology, 22(9), 1026.
Thring, R. H. (1989). Homogeneous charge compression ignition (HCCI) engines. SAE
Technical Paper Series, Paper No. 892068
Tobias, H. J., Beving, D. E., Ziemann, P. J., Sakurai, H., Zuk, M., McMurry, P. H., et
al. (2001). Chemical analysis of diesel engine nanoparticles using a nanoDMA/Thermal desorption particle beam mass spectrometer. Environmental
Science & Technology, 35(11), 2233-2243.
Tsolakis, A., & Megaritis, A. (2005). Partially premixed charge compression ignition
engine with on-board H2 production by exhaust gas fuel reforming of diesel and
biodiesel. International Journal of Hydrogen Energy, 30(7), 731-745.
Tsolakis, A., Megaritis, A., & Yap, D. (2008). Application of exhaust gas fuel
reforming in diesel and homogeneous charge compression ignition (HCCI) engines
fuelled with biofuels. Energy, 33(3), 462-470.
Vaaraslahti, K., Keskinen, J., Giechaskiel, B., Solla, A., Murtonen, T., & Vesala, H.
(2005). Effect of lubricant on the formation of heavy-duty diesel exhaust
nanoparticles. Environmental Science Technology, 39(21), 8497.
Wallner, T., & Frazee, R. (2010). Study of regulated and non-regulated emissions from
combustion of gasoline, alcohol fuels, and their blends in a DI-SI engine. SAE
Technical Paper Series, Paper No. 2010-01-1571
Wang, S. C., & Flagan, R. C. (1990). Scanning electrical mobility spectrometer.
Aerosol Science and Technology, 13(2), 230-240.
Weingartner, E., Baltensperger, U., & Burtscher, H. (1995). Growth and structural
change of combustion aerosols at high relative humidity. Environmental Science &
Technology, 29(12), 2982-2986.
183
Westbrook, C. K. (2000). Chemical kinetics of hydrocarbon ignition in practical
combustion systems. Proceedings of the Combustion Institute, 28, 1563-1577.
Westbrook, C. K., & Dryer, F. L. (1984). Chemical kinetic modeling of hydrocarbon
combustion. Progress in Energy and Combustion Science, 10(1), 1-57.
Wiedensohler, A., Orsini, D., Covert, D. S., Coffmann, D., Cantrell, W., Havlicek, M.,
et al. (1997). Intercomparison study of the size-dependent counting efficiency of 26
condensation particle counters. Aerosol Science & Technology, 27(2), 224-242.
Woschni, G. (1967). A universally applicable equation for the instantaneous heat
transfer coefficient in the internal combustion engine. SAE Technical Paper Series,
Paper No. 670931
Yang, J. (2005). Expanding the operating range of homogeneous charge compression
ignition–spark ignition dual-mode engines in the homogeneous charge compression
ignition mode. International Journal of Engine Research, 6(4), 279-288.
Yap, D., Megaritis, A., Peucheret, S., & Wyszynski, M. L. (2004). Effect of hydrogen
addition on natural gas HCCI combustion. SAE Technical Paper Series, Paper No.
2004-01-1972
Yap, D., Megaritis, A., & Wyszynski, M. L. (2004). An investigation into bioethanol
homogeneous charge compression ignition (HCCI) engine operation with residual
gas trapping. Energy & Fuels, 18(5), 1315-1323.
Yap, D., Peucheret, S. M., Megaritis, A., Wyszynski, M. L., & Xu, H. (2006). Natural
gas HCCI engine operation with exhaust gas fuel reforming. International Journal
of Hydrogen Energy, 31(5), 587-595.
Yilmaz, E., Tian, T., Wong, V. W., & Heywood, J. B. (2002). An experimental and
theoretical study of the contribution of oil evaporation to oil consumption. SAE
Technical Paper Series, Paper No. 2002-01-2684
Yilmaz, E., Tian, T., Wong, V. W., & Heywood, J. B. (2004). The contribution of
different oil consumption sources to total oil consumption in a spark ignition
engine. SAE Technical Paper Series, Paper No. 2004-01-2909
Zhao, F., Lai, M., & Harrington, D. L. (1995). The spray characteristics of automotive
port fuel injection- a critical review. SAE Technical Paper Series, Paper No.
950506
Zheng, M., Reader, G. T., & Hawley, J. G. (2004). Diesel engine exhaust gas
recirculation––a review on advanced and novel concepts. Energy Conversion and
Management, 45(6), 883-900.
184
Zinola, S., & Levy, J. (2009). Detailed particulate characterization from HCCI
combustion for future DPF development. SAE Technical Paper Series,
185